Upload
ngokhanh
View
229
Download
0
Embed Size (px)
Citation preview
TECHNICAL REPORT
© The Broadband Forum. All rights reserved.
TR-350
Ethernet Services using BGP MPLS Based Ethernet VPNs (EVPN)
Issue: 2
Issue Date: TBD 2017
1
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 2 of 65
Notice 2 3 The Broadband Forum is a non-profit corporation organized to create guidelines for broadband 4
network system development and deployment. This Broadband Forum Technical Report has been 5
approved by members of the Forum. This Broadband Forum Technical Report is not binding on 6
the Broadband Forum, any of its members, or any developer or service provider. This Broadband 7
Forum Technical Report is subject to change, but only with approval of members of the Forum. 8
This Technical Report is copyrighted by the Broadband Forum, and all rights are reserved. 9
Portions of this Technical Report may be copyrighted by Broadband Forum members. 10
11
THIS SPECIFICATION IS BEING OFFERED WITHOUT ANY WARRANTY WHATSOEVER, 12
AND IN PARTICULAR, ANY WARRANTY OF NONINFRINGEMENT IS EXPRESSLY 13
DISCLAIMED. ANY USE OF THIS SPECIFICATION SHALL BE MADE ENTIRELY AT THE 14
IMPLEMENTER'S OWN RISK, AND NEITHER the Forum, NOR ANY OF ITS MEMBERS OR 15
SUBMITTERS, SHALL HAVE ANY LIABILITY WHATSOEVER TO ANY IMPLEMENTER 16
OR THIRD PARTY FOR ANY DAMAGES OF ANY NATURE WHATSOEVER, DIRECTLY 17
OR INDIRECTLY, ARISING FROM THE USE OF THIS SPECIFICATION. 18
19
Broadband Forum Technical Reports may be copied, downloaded, stored on a server or otherwise 20
re-distributed in their entirety only, and may not be modified without the advance written 21
permission of the Broadband Forum. 22
23
The text of this notice must be included in all copies of this Broadband Forum Technical Report 24
25
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 3 of 65
26
Issue Number Approval
Date
Publication Date Issue Editor Changes
1 9 Nov. 2015 11 Nov. 2015 Rao Cherukuri,
Juniper
Sriganesh Kini,
Ericsson
Original
bbf2016.691.00 TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Steven Blake,
Ericsson
Added Sec.
13 EPL and
EVPL
bbf2016.691.01 TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Steven Blake,
Ericsson
Correct [51]
URL, R-81;
fix editorial
nits.
bbf2016.691.02 TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Steven Blake,
Ericsson
Same as .01
with all
changes
accepted. WT-350 ELine Modifications from bbf2016.691.02-Cherukuri-DF-bbf2016.691.doc
TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Steven Blake,
Ericsson
Same as .02
with changes
for
Bandwidth
Policy and
updated
figures for
Eline. WT-350 ELine final Modifications from bbf2016.691.02-Cherukuri-DF-bbf2016.691.docx
TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Same as
above with
changes for
ETree.
WT-350 ELine
final
Modifications
from
bbf2016.691.02
-Cherukuri-DF-
bbf2016.691.do
cx
TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
Proposed SB
Candidate
WT-
350i2ELineETr
TBD TBD Rao Cherukuri, BBF
Distinguished Fellow
SB Text
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 4 of 65
ee.SBCandidate
-Sinicrope-
RnTWAD
28
29
30
31
Comments or questions about this Broadband Forum Technical Report should be directed to 32
34
35
Editor Rao Cherukuri BBF Distinguished Fellow
Steven Blake Ericsson
Routing and Transport
WA Director
David Sinicrope
Ericsson
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 5 of 65
TABLE OF CONTENTS 36
EXECUTIVE SUMMARY ........................................................................................................... 10 37
1 PURPOSE AND SCOPE ....................................................................................................... 11 38
PURPOSE .............................................................................................................................. 11 39
SCOPE .................................................................................................................................. 11 40
2 REFERENCES AND TERMINOLOGY ............................................................................. 13 41
CONVENTIONS ..................................................................................................................... 13 42
REFERENCES ........................................................................................................................ 13 43
DEFINITIONS ........................................................................................................................ 16 44
ABBREVIATIONS .................................................................................................................. 17 45
3 TECHNICAL REPORT IMPACT ...................................................................................... 19 46
ENERGY EFFICIENCY ............................................................................................................ 19 47
IPV6 ..................................................................................................................................... 19 48
SECURITY ............................................................................................................................. 19 49
PRIVACY .............................................................................................................................. 19 50
4 CARRIER ETHERNET SERVICES ................................................................................... 20 51
CARRIER ETHERNET REQUIREMENTS ................................................................................... 20 52
5 LAYER 2 ETHERNET VPNS IN MPLS NETWORKS .................................................... 21 53
6 REFERENCE ARCHITECTURE ....................................................................................... 22 54
GENERAL REFERENCE ARCHITECTURE ................................................................................ 22 55
MPLS FOR CARRIER ETHERNET IN BROADBAND ACCESS & AGGREGATION ....................... 23 56
6.2.1 Multi-Service Broadband Access & Aggregation .................................................... 23 57
6.2.2 TR-178 Architectures ............................................................................................... 23 58
7 SIGNALING AND ROUTING ............................................................................................. 24 59
LSP SIGNALING ................................................................................................................... 24 60
7.1.1 Multi-area LSP Signaling ........................................................................................ 24 61
ROUTING .............................................................................................................................. 26 62
8 OAM ........................................................................................................................................ 27 63
ETHERNET OAM.................................................................................................................. 27 64
8.1.1 Link OAM ................................................................................................................. 27 65
8.1.2 ITU-T G.8013/Y.1731 .............................................................................................. 28 66
MEF SERVICE OAM ............................................................................................................ 28 67
MPLS OAM ........................................................................................................................ 29 68
8.3.1 LSP OAM ................................................................................................................. 29 69
8.3.2 Convergence ............................................................................................................ 30 70
9 QOS ......................................................................................................................................... 32 71
TUNNEL COS MAPPING AND MARKING ................................................................................. 32 72
10 PSN RESILIENCY ........................................................................................................ 33 73
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 6 of 65
FAILURE DETECTION ............................................................................................................ 33 74
LSP RECOVERY .................................................................................................................... 33 75
CONTROL PLANE RESILIENCY ............................................................................................... 34 76
11 BGP MPLS-BASED ETHERNET VPN ...................................................................... 35 77
REFERENCE ARCHITECTURE AND OVERVIEW ...................................................................... 35 78
EVPN SERVICE INTERFACES ............................................................................................... 36 79
11.2.1 VLAN-Based Service Interfaces ............................................................................... 36 80
11.2.2 VLAN Bundle Service Interfaces .............................................................................. 36 81
11.2.3 VLAN-Aware Bundle Service Interfaces .................................................................. 37 82
DATA PLANE ........................................................................................................................ 37 83
11.3.1 Underlying PSN transport ....................................................................................... 37 84
11.3.2 VPN encapsulation................................................................................................... 37 85
11.3.3 VID Translation ....................................................................................................... 37 86
11.3.4 Frame Ordering ....................................................................................................... 37 87
CONTROL PLANE.................................................................................................................. 37 88
MULTI-HOMING AND LOAD BALANCING .............................................................................. 38 89
11.5.1 All-Active Redundancy Mode ................................................................................... 38 90
11.5.2 Single-Active Redundancy Mode ............................................................................. 38 91
FAST CONVERGENCE ........................................................................................................... 39 92
12 EVPN ENABLED MULTIPOINT TO MULTIPOINT ETHERNET VPN 93
SERVICES (E-LAN) ..................................................................................................................... 40 94
ETHERNET PRIVATE LAN (EP-LAN) .................................................................................. 40 95
ETHERNET VIRTUAL PRIVATE LAN (EVP-LAN) ................................................................ 41 96
EVPN FOR ESTABLISHING EP-LAN AND EVP-LAN ........................................................... 41 97
12.3.1 Service Interfaces ..................................................................................................... 42 98
12.3.2 Data plane ................................................................................................................ 43 99
12.3.3 Tunnel signaling....................................................................................................... 43 100
12.3.4 Routing ..................................................................................................................... 43 101
12.3.5 Multi-Homing and Load balancing ......................................................................... 43 102
12.3.6 OAM ......................................................................................................................... 44 103
12.3.7 Convergence ............................................................................................................ 44 104
12.3.8 PSN Resiliency ......................................................................................................... 44 105
12.3.9 Multicast and Broadcast .......................................................................................... 44 106
12.3.10 QoS ....................................................................................................................... 44 107
12.3.11 Security ................................................................................................................ 44 108
SUPPORT OF SERVICE ATTRIBUTES FOR EP-LAN AND EVP-LAN ........................................ 45 109
12.4.1 Bandwidth Profile .................................................................................................... 45 110
12.4.2 Bundling ................................................................................................................... 45 111
12.4.3 CE-VLAN ID preservation for EVC ......................................................................... 45 112
12.4.4 CE-VLAN CoS preservation for EVC ...................................................................... 46 113
12.4.5 EVC Maximum Service Frame Size ......................................................................... 46 114
12.4.6 Frame delivery ......................................................................................................... 46 115
12.4.7 Layer 2 control protocols......................................................................................... 47 116
12.4.8 EVC performance..................................................................................................... 47 117
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 7 of 65
13 EVPN ENABLED POINT TO POINT ETHERNET VPN SERVICES (E-LINE) . 49 118
ETHERNET PRIVATE LINE (EPL) .......................................................................................... 49 119
ETHERNET VIRTUAL PRIVATE LINE (EVPL) ........................................................................ 50 120
EVPN FOR ESTABLISHING EPL AND EVPL ......................................................................... 51 121
13.3.1 Service Interfaces ..................................................................................................... 51 122
13.3.2 Operation ................................................................................................................. 52 123
13.3.3 Data Plane ............................................................................................................... 52 124
13.3.4 BGP extensions ........................................................................................................ 52 125
13.3.5 Tunnel signaling....................................................................................................... 53 126
13.3.6 Routing ..................................................................................................................... 53 127
13.3.7 Multi-Homing and Load balancing ......................................................................... 53 128
13.3.8 OAM ......................................................................................................................... 53 129
13.3.9 Convergence ............................................................................................................ 53 130
13.3.10 PSN Resiliency ..................................................................................................... 53 131
13.3.11 QoS ....................................................................................................................... 53 132
13.3.12 Security ................................................................................................................ 54 133
SUPPORT OF SERVICE ATTRIBUTES FOR EPL AND EVPL ...................................................... 54 134
13.4.1 Bandwidth Profile .................................................................................................... 54 135
13.4.2 Bundling ................................................................................................................... 54 136
13.4.3 CE-VLAN ID preservation for EVC ......................................................................... 55 137
13.4.4 CE-VLAN CoS preservation for EVC ...................................................................... 55 138
13.4.5 EVC Maximum Service Frame Size ......................................................................... 55 139
13.4.6 Frame delivery ......................................................................................................... 56 140
13.4.7 Layer 2 control protocols......................................................................................... 56 141
13.4.8 EVC performance..................................................................................................... 56 142
13.4.9 Multiple Class of Service ......................................................................................... 57 143
14 EVPN ENABLED ROOTED-MULTIPOINT ETHERNET VPN SERVICES (E-144
TREE) 58 145
ETHERNET PRIVATE TREE (EP-TREE) .................................................................................. 58 146
ETHERNET VIRTUAL PRIVATE TREE (EVP-TREE) ................................................................ 59 147
EVPN FOR ESTABLISHING EP-TREE AND EVP-TREE ........................................................... 60 148
14.3.1 E-Tree support in EVPN .......................................................................................... 61 149
14.3.2 Service Interfaces ..................................................................................................... 61 150
14.3.3 Data plane ................................................................................................................ 62 151
14.3.4 Tunnel signaling....................................................................................................... 62 152
14.3.5 Routing ..................................................................................................................... 62 153
14.3.6 Multi-Homing and Load balancing ......................................................................... 62 154
14.3.7 OAM ......................................................................................................................... 62 155
14.3.8 Convergence ............................................................................................................ 62 156
14.3.9 PSN Resiliency ......................................................................................................... 62 157
14.3.10 Multicast and Broadcast ...................................................................................... 62 158
14.3.11 QoS ....................................................................................................................... 63 159
14.3.12 Security ................................................................................................................ 63 160
SUPPORT OF SERVICE ATTRIBUTES FOR EP-TREE AND EVP-TREE ....................................... 63 161
14.4.1 Bandwidth Profile .................................................................................................... 63 162
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 8 of 65
14.4.2 Bundling ................................................................................................................... 64 163
14.4.3 CE-VLAN ID preservation for EVC ......................................................................... 64 164
14.4.4 CE-VLAN CoS preservation for EVC ...................................................................... 64 165
14.4.5 EVC Maximum Service Frame Size ......................................................................... 64 166
14.4.6 Frame delivery ......................................................................................................... 65 167
14.4.7 Layer 2 control protocols......................................................................................... 65 168
14.4.8 EVC performance..................................................................................................... 65 169 170
171
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 9 of 65
List of Figures 172
173
Figure 1 Reference Architecture ...................................................................................................... 22 174
Figure 2 Components of OAM ........................................................................................................ 27 175
Figure 3 EVPN Architecture for Ethernet services using BGP MPLS ............................................ 36 176
Figure 4 Ethernet Private LAN (EP-LAN) Service ......................................................................... 40 177
Figure 5 Ethernet Virtual Private LAN (EVP-LAN) Service .......................................................... 41 178
Figure 6 Ethernet Private Line (EPL) Service ................................................................................. 50 179
Figure 7 Ethernet Virtual Private Line (EVPL) Service .................................................................. 51 180
Figure 8 Ethernet Private Tree (EP-Tree) Service ........................................................................... 59 181
Figure 9 Ethernet Virtual Private Tree (EVP-Tree) Service ............................................................ 60 182
183
184
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 10 of 65
Executive Summary 185
Carrier Ethernet provides extensions to Ethernet, enabling telecommunications network providers 186
to provide Ethernet services to customers and to utilize Ethernet technology in their networks. 187
188
Carrier Ethernet services are being used in Broadband access networks, enterprise networks and 189
backhaul networks. Providing Carrier Ethernet services using MPLS network infrastructure is 190
generating revenue opportunities for global carriers, driven by customer demand for higher 191
bandwidth connectivity. Though TR-224 describes the architecture for solutions to implement 192
Carrier Ethernet services using an MPLS network, the VPLS based solution has a number of 193
limitations when it comes to redundancy, multicast optimization and provisioning simplicity. It 194
does not address requirements such as multi-homing with all-active forwarding, load balancing, 195
policy-based control and control plane based MAC learning. Service interface requirements for 196
data-center interconnects are also not addressed by TR-224. 197
198
This document provides technical architecture and equipment requirements to implement the 199
Carrier Ethernet services using BGP MPLS-based EVPNs in order to overcome the limitations of 200
VPLS and address the additional requirements. By specifying a common technical architecture, 201
common equipment requirements and common set of feature options, this document promotes 202
multi-vendor interoperability. 203
204
205
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 11 of 65
1 Purpose and Scope 206
Purpose 207
Carrier Ethernet provides extensions to Ethernet enabling telecommunications network providers 208
to provide Ethernet services to customers and to utilize Ethernet technology in their networks. 209
Service providers are deploying Carrier Ethernet services around the globe, in large part, because 210
Carrier Ethernet has compelling capabilities such as standardized service definitions as well as 211
improved scalability, reliability, QoS, and manageability. 212
213
Carrier Ethernet services are being used in Broadband access networks, enterprise networks and 214
backhaul networks. The integration of Ethernet into MPLS network infrastructure is generating 215
revenue opportunities for global carriers, driven by customer demand for higher bandwidth 216
connectivity. This document provides a technical architecture and equipment requirements 217
implementing the specified Ethernet services using BGP MPLS-based Ethernet VPNs (EVPNs) in 218
IP/MPLS network. 219
220
New Ethernet service applications require capabilities such as: multi-homing with all-active 221
forwarding; load balancing; policy based control, and control plane MAC learning. TR-224 [4] and 222
TR-178 [2] based solutions do not provide these features; solutions based on BGP MPLS EVPNs 223
do. 224
225
By specifying a common technical architecture, common equipment requirements and common set 226
of feature options, this document promotes multi-vendor interoperability. This document may be 227
used as a basis for conformance testing. 228
Scope 229
This document defines a reference architecture for Carrier Ethernet services using BGP MPLS-230
based Ethernet VPN mechanisms: 231
232
• Ethernet multipoint to multipoint (E-LAN) 233
• Ethernet point to point (E-Line) 234
• Ethernet point to multipoint (E-Tree) including E-Tree* 235
• Ethernet access to support wholesale access service 236
• Control, OAM, QoS, reliability and scalability for the MPLS network 237
• Support Ethernet service capabilities specified in RFC 7209 [37] 238
239
This document specifies how to implement the Ethernet services layer. It does not specify the 240
service layer itself. Ethernet Control and OAM protocols will be transparently transported, except 241
for cases where Layer 2 control protocol processing is required per-service definition. 242
243
This document includes the Carrier Ethernet E-LAN, E-Line, and E-Tree (including E-Tree*) 244
services types using BGP MPLS EVPNs. 245
246
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 12 of 65
247
In order to support Carrier Ethernet services across multiple networks, the scope of this document 248
includes the following: 249
• Attachment circuits (AC) providing a user-to-network interface complying with the MEF 250
UNI are supported. 251
• Ethernet attachment circuits for multi-service broadband access and aggregation (i.e., TR-252
101/TR-178) are supported. 253
• Additional Ethernet service capabilities of BGP MPLS-based EVPNs (e.g. multi-homing 254
with all-active forwarding, load balancing, policy based control, control based MAC 255
learning, etc) are supported. 256
• Support for interworking with TR-224. 257
• To support Carrier Ethernet across multiple SP networks, the specification addresses multi 258
autonomous systems which preserves end-to-end capabilities (e.g., OAM, QoS and 259
protection etc). 260
• Cases where the UNI-N functions are or are not collocated with the PE are supported. 261
262
TR-350 provides technical architecture and equipment requirements implementing MEF Carrier 263
Ethernet services with BGP MPLS EVPNs. EVPNs architecture and protocols are based on 264
BGP/MPLS IP VPNs , which supports multiple domains. This capability is used to support 265
connectivity between service endpoints (e.g. MEF UNIs) connected to different networks or 266
operators. 267
TR-350 does not use architecture or connectivity models of Carrier Ethernet using MEF 26.1 [46]. 268
269
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 13 of 65
2 References and Terminology 270
Conventions 271
In this Technical Report, several words are used to signify the requirements of the specification. 272
These words are always capitalized. More information can be found be in RFC 2119 [8]. 273
274
MUST This word, or the term “REQUIRED”, means that the definition is an
absolute requirement of the specification.
MUST NOT This phrase means that the definition is an absolute prohibition of the
specification.
SHOULD This word, or the adjective “RECOMMENDED”, means that there
could exist valid reasons in particular circumstances to ignore this
item, but the full implications need to be understood and carefully
weighed before choosing a different course.
SHOULD NOT This phrase, or the phrase "NOT RECOMMENDED" means that there
could exist valid reasons in particular circumstances when the
particular behavior is acceptable or even useful, but the full
implications need to be understood and the case carefully weighed
before implementing any behavior described with this label.
MAY This word, or the adjective “OPTIONAL”, means that this item is one
of an allowed set of alternatives. An implementation that does not
include this option MUST be prepared to inter-operate with another
implementation that does include the option.
References 275
The following references are of relevance to this Technical Report. At the time of publication, the 276
editions indicated were valid. All references are subject to revision; users of this Technical Report 277
are therefore encouraged to investigate the possibility of applying the most recent edition of the 278
references listed below. 279
A list of currently valid Broadband Forum Technical Reports is published at 280
www.broadband-forum.org. 281
282
Document Title Source Year
TR-101 Migration to Ethernet-Based Broadband
Aggregation
BBF 2011
TR-178 Multi-service Broadband Network
Architecture and Nodal Requirements
BBF 2014
TR-221 Technical Specifications for MPLS in Mobile
Backhaul Networks
BBF 2011
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 14 of 65
TR-224 Technical Specification for MPLS in Carrier
Ethernet Networks
BBF 2014
IEEE 802.3 IEEE Standard Ethernet IEEE 2012
IEEE 802.1Q IEEE Standard for Local and metropolitan
area networks--Media Access Control (MAC)
Bridges and Virtual Bridged Local Area
Networks
IEEE 2011
RFC 1195 Use of OSI IS-IS for Routing in TCP/IP and
Dual Environments
IETF 1990
RFC 2119 Key words for use in RFCs to Indicate
Requirement Levels
IETF 1997
RFC 2328 OSPF Version 2 IETF 1998
RFC 3209 RSVP-TE: Extensions to RSVP for LSP
Tunnels
IETF 2001
RFC 3270 Multi-Protocol Label Switching (MPLS)
Support of Differentiated Services
IETF 2002
RFC 3473 Generalized Multi-Protocol Label Switching
(GMPLS) Signaling Resource ReserVation
Protocol-Traffic Engineering (RSVP-TE)
Extensions
IETF 2003
RFC 3478 Graceful Restart Mechanism for Label
Distribution Protocol
IETF 2003
RFC 3564 Requirements for Support of Differentiated
Services-aware MPLS Traffic Engineering
IETF 2003
RFC 3623 Graceful OSPF Restart IETF 2003
RFC 3630 Traffic Engineering (TE) Extensions to OSPF
Version 2
IETF 2003
RFC 5306 Restart Signaling for Intermediate System to
Intermediate System (IS-IS)
IETF 2004
RFC 4090 Fast Reroute Extensions to RSVP-TE for LSP
Tunnels
IETF 2005
RFC 4124 Protocol Extensions for Support of Diffserv-
aware MPLS Traffic Engineering
IETF 2005
RFC 4206 Label Switched Paths (LSP) Hierarchy with
Generalized Multi-Protocol Label Switching
(GMPLS) Traffic Engineering (TE)
IETF 2005
RFC 4364 BGP/MPLS IP Virtual Private Networks
(VPNs)
IETF 2006
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 15 of 65
RFC 4379 Detecting Multi-Protocol Label Switched
(MPLS) Data Plane Failures
IETF 2006
RFC 4761 Virtual Private LAN Service (VPLS) Using
BGP for Auto-Discovery and Signaling
IETF 2007
RFC 4762 Virtual Private LAN Service (VPLS) Using
BGP for Auto-Discovery and Signaling
IETF 2007
RFC 5036 LDP Specification IETF 2007
RFC 5150 Label Switched Path Stitching with
Generalized Multiprotocol Label Switching
Traffic Engineering (GMPLS TE)
IETF 2008
RFC 5151 Inter-Domain MPLS and GMPLS Traffic
Engineering -- Resource Reservation
Protocol-Traffic Engineering (RSVP-TE)
Extensions
IETF 2008
RFC 5283 LDP Extension for Inter-Area Label Switched
Paths (LSPs)
IETF 2008
RFC 5286 Basic Specification for IP Fast Reroute:
Loop-Free Alternates
IETF 2008
RFC 5305 IS-IS Extensions for Traffic Engineering IETF 2008
RFC 5586 MPLS Generic Associated Channel IEIF 2009
RFC 5880 Bidirectional Forwarding Detection (BFD) IETF 2010
RFC 5881 Bidirectional Forwarding Detection (BFD)
for IPv4 and IPv6 (Single Hop)
IETF 2010
RFC 5884 Bidirectional Forwarding Detection (BFD)
for MPLS Label Switched Paths (LSPs)
IETF 2010
RFC 6424 Mechanism for Performing Label Switched
Path Ping (LSP Ping) over MPLS Tunnels
IETF 2011
RFC 6790 The Use of Entropy Labels in MPLS
Forwarding
IETF 2012
RFC 7209 Requirements for Ethernet VPN (EVPN) IETF 2014
RFC 7387 Framework for Ethernet Tree (E-Tree)
Service over a Multiprotocol Label Switching
(MPLS) Network
IETF 2014
RFC 7432 BGP MPLS Based Ethernet VPN IETF 2015
G.8013/Y.1731 OAM functions and mechanisms for Ethernet
based networks
ITU-T 2013
MEF 6.1 Ethernet Services Definitions - Phase 2 MEF 2008
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 16 of 65
MEF 10.2 Ethernet Services Attributes - Phase 2 MEF 2009
MEF 30 Service OAM Fault Management
Implementation Agreement
MEF 2011
MEF 22.1 Mobile Backhaul Phase 2 Implementation
Agreement
MEF 2012
MEF 23.1 Carrier Ethernet Class of Service – Phase 2 MEF 2012
MEF 26.1 External Network Network Interface (ENNI) –
Phase 2
MEF 2012
MEF 35 Service OAM Performance Monitoring
Implementation Agreement
MEF 2012
MEF 6.1.1 Layer 2 Control Protocol Handling
Amendment to MEF 6.1
MEF 2012
MEF 10.3 Ethernet Services Attributes - Phase 3 MEF 2013
MEF 6.2 EVC Ethernet Services Definitions - Phase 3 MEF 2014
MEF 45 Multi-CEN L2CP MEF 2014
draft-ietf-bess-
evpn-etree
E-TREE Support in EVPN & PBB-EVPN IETF 2017
draft-ietf-bess-
evpn-vpws
Virtual Private Wire Support in EVPN IETF 2017
Definitions 283
The following terminology is used throughout this Technical Report. 284
285
AGN An aggregation node (AGN) is a node which aggregates several access nodes
(ANs).
AN An access node is a node which processes customer frames or packets at
Layer 2 or above. This includes but is not limited to DSLAMs or OLTs (in
case of (G)PON deployments).
E-Line A service connecting two customer Ethernet ports over a WAN.
E-LAN A multipoint service connecting a set of customer endpoints, giving the
appearance to the customer of a bridged Ethernet network connecting the
sites.
E-Tree A rooted multipoint Ethernet Virtual Connection. The E-Tree service type
can support one or multiple Root UNIs (see Section 9.3/MEF 6.2 [50])
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 17 of 65
E-Tree* Partially implementing MEF multipoint service connecting only one root and
a set of leaves, but preventing inter-leaf communication. See details in TR-
221.
Note: Ethernet Tree (E-Tree) service type is specified in Section 6.3/MEF 6.1
[41]. The Appendix in TR-221 modifies the E-Tree service type which is
used in different services. The modified E-Tree* service type is used in both
Ethernet Private Tree service and Ethernet Virtual Private Tree Service
specified in Section 14 .
Abbreviations 286
This Technical Report uses the following abbreviations: 287
288
AC Attachment Circuit
AGN Aggregation Node
AN Access Node
BFD Bidirectional Forwarding Detection
BGP Border Gateway Protocol
BNG Broadband Network Gateway
CE Customer Edge
CoS Class of Service
CV Connectivity Verification
EPL Ethernet Private Line
EP-LAN Ethernet Private-LAN
EVC Ethernet Virtual Connection
EVPL Ethernet Virtual Private Line
EVP-LAN Ethernet Virtual Private – LAN
FD Frame Delay
FRR Fast Re-route
FLR Frame Loss Ratio
H-VPLS Hierarchal Virtual Private LAN Service
IETF Internet Engineering Task Force
IP Internet Protocol
ITU-T International Telecommunication Union Telecommunication Standardization
Sector
L2VPN Layer 2 Virtual Private Network
LAN Local Area Network
LER Label Edge Router
LFA Loop Free Alternate
LSP Label Switched Path
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 18 of 65
LSR Label Switch Router
MAC Medium Access Control
MEF Metro Ethernet Forum
MPLS Multi Protocol Label Switching
OAM Operations, Administration and Management
OAMPDU OAM Protocol Data Unit
P Provider
PE Provider Edge
PSN Packet Switched Network
PW Pseudowire
QoS Quality of Service
RFC Request for Comments
RSVP-TE Resource ReSerVation Protocol with Traffic Engineering extensions
SLA Service Level Agreement
TE Traffic Engineering
T-LDP Targeted Label Distribution Protocol
TLV Type/Length/Value
TR Technical Report
UNI User to Network Interface
UDP User Datagram Protocol
VPLS Virtual Private LAN Service
VPN Virtual Private Network
VPWS Virtual Private Wire Service
WG Working Group
289
290
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 19 of 65
3 Technical Report Impact 291
Energy Efficiency 292
TR-350 has no impact on energy efficiency. 293
IPv6 294
Carrier Ethernet services operate at Layer 2 and therefore the network is agnostic to IPv6 user 295
traffic. The IPv6 header DSCP field is assumed to be mapped to the Ethernet P bits by the service 296
user. 297
298
IPv6 addressing may appear in its respective places in control, OAM, and management protocols, 299
e.g., node ids, FECs, and loopback addresses, etc. 300
301
TR-350 has no direct impact on IPv6. 302
Security 303
Security requirements are specified for each service in respective sections. 304
Privacy 305
Any issues regarding privacy are not affected by TR-350. 306
307
308
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 20 of 65
4 Carrier Ethernet services 309
Ethernet is now being used as both a transport technology and a service delivery architecture. The 310
MEF Carrier Ethernet specifies Ethernet service type, service attributes, QoS and SLA. The 311
service type includes point-to-point (E-Line), point-to-multipoint (E-Tree), multipoint-to-312
multipoint (E-LAN), and E-Access. The service definition includes both port-based and VLAN-313
based service identification. 314
315
TR-224 refers to MEF 6.1 and MEF 10.2. TR-350 uses the backward compatible subset of the 316
revised specifications MEF 6.2 [50] and MEF 10.3 [49] to achieve the equivalent function. This 317
makes WT-350 compatible with TR-224. 318
319
The MEF also defined Carrier Ethernet as a ubiquitous, standardized, carrier-class service and 320
network defined by attributes that distinguish Carrier Ethernet from familiar LAN-based Ethernet. 321
Carrier Ethernet Requirements 322
Service providers worldwide are migrating their existing networks to deliver Carrier Ethernet 323
services to enterprises, businesses & residential end-users. The attributes are as follows: 324
325
1. Standardized Services 326
• Support E-Line, E-LAN and E-Tree service types as defined by MEF 327
• No changes to customer LAN equipment or networks and accommodates existing 328
network connectivity such as, time-sensitive, TDM traffic and signaling 329
• Wide choice and granularity of bandwidth and quality of service options 330
2. Security 331
3. Scalability 332
• The ability to support millions of Ethernet Virtual Connection (EVC) services for 333
enterprise and residential users 334
• Scalability of bandwidth from 1 Mbps to 10 Gbps and beyond, in granular 335
increments 336
4. Reliability 337
• The ability for the network to detect & recover from faults quickly 338
• Fast network convergence 339
5. Quality of Service 340
• Service Level Agreements (SLAs) that deliver end-to-end performance 341
• Traffic profile enforcement per-EVC 342
• Hierarchical queuing 343
6. Service Management 344
• Minimize network touch points in provisioning 345
• Standards-based OAM to support SLA 346
347
348
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 21 of 65
5 Layer 2 Ethernet VPNs in MPLS Networks 349
MPLS has for a long time been defined as a convergence technology, one that will allow service 350
providers to bring together their disparate networks and leverage features like traffic engineering, 351
hierarchal QoS, and service interworking. 352
353
Provider Provisioned Virtual Private Networks (PPVPN) now dominates the IP-VPN services 354
market and is projected for significant growth. Many service providers have provided Ethernet 355
VPN services using Virtual Private LAN Services (VPLS) as an alternative that allows enterprises 356
to manage their own routing. 357
358
TR-224 uses VPLS to support Ethernet LAN services in IP/MPLS networks. A VPLS PE 359
emulates an Ethernet bridge (IEEE 802.1Q [6]) and performs MAC learning in the data plane. 360
New applications using Ethernet services require capabilities such as: multi-homing with all-active 361
forwarding, load balancing, policy based control, and control plane MAC learning. To support 362
these capabilities IETF developed BGP MPLS-based Ethernet VPNs (EVPNs). TR-224 based 363
solutions do not provide these features. 364
365
When an Ethernet multipoint service is provided using EVPN, control-plane-based remote MAC 366
learning is used over the MPLS core (PE to PE) network. MAC learning between PE and CE is 367
done in the data plane. EVPN is designed to handle multi-homing, and per-flow load balancing. 368
The EVPN technology uses MP-BGP over an MPLS network. The technology is similar to BGP 369
MPLS-based IP VPNs (RFC 4364). Using MP-BGP to distribute the reachability of MAC 370
addresses over a MPLS network brings the same operational control and scale of L3VPN to 371
L2VPN. 372
373
The EVPN solution provides a common base for all Ethernet service types including E-LAN, E-374
Line, E-Tree (including E-Tree* from TR-221), E-Access and enables these services to be created 375
such that they can span across domains. In addition to the common base above, BGP MPLS-based 376
EVPNs also provide solutions for the requirements in RFC 7209 [37] including: 377
• Multi-homing: with all active forwarding and load balancing from CE to CE. VPLS can 378
only support multi-homing with single active mode. 379
• Flow based load balancing and multipath 380
• Multicast optimization: must be able to support P2MP MPLS LSPs and MP2MP MPLS 381
LSPs. VPLS only supports P2MP MPLS LSPs. 382
• Fast convergence to minimize downtime and packet loss. 383
• Support MAC mobility to support cloud services. 384
385
386
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 22 of 65
6 Reference Architecture 387
General Reference Architecture 388
Figure 1 provides a generic overview of how Carrier Ethernet Services can be deployed using a 389
BGP MPLS-based EVPN infrastructure, including basic reference points and their functional roles. 390
Depending on the application, non-MEF-defined Ethernet Attachment Circuits and Attachment 391
Circuits providing User-to-Network interfaces complying with Metro Ethernet Forum definitions 392
(MEF UNI) are supported. Multi-domain connectivity and external handoff are also supported. 393
394
395 Figure 1 Reference Architecture 396
397
Defined as business interfaces supporting the service handoff between different parties (between 398
user and provider or between providers, respectively), UNI has two functions: 399
400
1. provide reference points for network demarcation 401
2. provide associated functionality 402
403
For deploying Metro Ethernet Forum-compliant Ethernet Services over MPLS, PE nodes need to 404
support the corresponding MEF UNI functionality at Attachment Circuit interfaces. 405
406
407
408
409
410
411
412
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 23 of 65
MPLS for Carrier Ethernet in Broadband Access & Aggregation 413
6.2.1 Multi-Service Broadband Access & Aggregation 414
For Multi-Service Broadband access and aggregation architecture see Section 6.2.1/TR-224 [4]. 415
6.2.2 TR-178 Architectures 416
There are two reference architectures that are being used to represent TR-178 networks: 1) MPLS-417
enabled access and 2) TR-101. For architectural details of MPLS-enabled access nodes and TR-418
101 see Section 6.2.2/TR-224 [4]. 419
420
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 24 of 65
7 Signaling and Routing 421
This section specifies the signaling protocol used to establish the underlying MPLS tunnel. Traffic 422
engineered PSN tunnels must be used when specific path (e.g. for protection purpose), QoS or 423
bandwidth constraints are required. 424
LSP Signaling 425
One of the following provisioning and signaling procedures are used for LSPs. 426
427
[R-1] PE and P routers supporting MPLS TE and non-TE LSPs MUST support one or 428
both of the following methods: 429
• Static provisioning 430
• Dynamic signaling 431
432
[R-2] Both of the following methods MUST be supported by PE and P routers for 433
dynamically signaled PSN tunnel LSPs: 434
• LDP is used to set up, maintain and release LSP tunnels per RFC 5036 [25]. 435
• RSVP-TE is used to set up, maintain and release LSPs for traffic engineered tunnels per 436
RFC 3209 [10] and RFC 5151 [27]. When traffic engineering is needed on the LSP, 437
RSVP-TE MUST be used. 438
439
[R-3] When co-routed bidirectional LSPs are required, GMPLS-RSVP-TE as per RFC 440
3473 [12] MAY be supported by PE and P routers. 441
7.1.1 Multi-area LSP Signaling 442
Several operators have multi-area networks for scalability. Link state Interior Gateway Protocols 443
(IGPs) such as OSPF (RFC 2328 [9]) and IS-IS (RFC 1195 [7]) allow dividing networks into areas 444
or levels so as to increase routing scalability within a routing domain. 445
446
Further some operators’ L2VPN networks span different geographical areas. To support these 447
networks, it is necessary to support inter-area and inter-AS (Autonomous System) Multiprotocol 448
Label Switching (MPLS) LSPs. 449
450
An “MPLS Domain” is considered to be any collection of network elements implementing MPLS 451
within a common realm of address space or path computation responsibility. Examples of such 452
domains include Autonomous Systems, Interior Gateway Protocol (IGP) routing areas, and 453
GMPLS overlay networks. 454
455
Signaling extensions for inter-area LSPs (that is, LSPs that traverse at least two IGP areas) are 456
required to ensure MPLS connectivity between PEs located in distinct IGP areas. 457
458
459
460
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 25 of 65
461
Multi-area RSVP-TE Signaling 462
Inter-domain TE LSPs can be supported by one of three options as specified in RFC 5151 [27] and 463
given below: 464
• contiguous LSPs 465
• nested LSPs 466
• stitched LSPs. 467
Note: While the specifications below cover both MPLS Traffic Engineering (TE) 468
and GMPLS, the intent is to use the TE coverage. 469
Contiguous 470
A contiguous TE LSP is a single TE LSP that is set up across multiple 471
domains using RSVP-TE signaling procedures described in Section 7.1. 472
473
Nested 474
One or more TE LSPs may be nested within another TE LSP as described in 475
RFC 4206 [20]. This technique can be used to nest one or more inter-476
domain TE LSPs into an intra-domain hierarchical LSP (H-LSP). The label 477
stacking construct is used to achieve nesting in packet networks. 478
479
To improve scalability, it may be useful to aggregate LSPs by creating a 480
hierarchy of such LSPs. 481
482
[R-4] PE routers SHOULD support establishment of RSVP-TE LSPs using 483
LSP hierarchy as per RFC 4206 [20]. 484
485
Stitched 486
LSP stitching signaling procedures are described in RFC 5150 [26]. This 487
technique can be used to stitch together shorter LSPs (LSP segments) to 488
create a single, longer LSP. The LSP segments of an inter-domain LSP may 489
be intra-domain LSPs or inter-domain LSPs. 490
491
The process of stitching LSP segments results in a single, end-to-end 492
contiguous LSP in the data plane. But in the control plane, each segment is 493
signaled as a separate LSP (with distinct RSVP sessions) and the end-to-end 494
LSP is signaled as yet another LSP with its own RSVP session. Thus, the 495
control plane operation for LSP stitching is very similar to that for nesting. 496
497
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 26 of 65
[R-5] PE routers SHOULD support establishment of RSVP-TE LSPs using 498
LSP stitching as per RFC 5150 [26]. 499
Multi-area LDP Signaling 500
RFC 5283 [28] facilitates the establishment of Label Switched Paths (LSPs) that would span 501
multiple IGP areas in a given Autonomous System (AS). 502
503
[R-6] PE routers SHOULD support establishment of inter-area LSPs using 504
LDP as per RFC 5283 [28]. 505
Routing 506
[R-7] One or both of the following methods MUST be supported by PE and P routers: 507
• Static routing 508
• Dynamic routing 509
510
[R-8] Both of the following methods MUST be supported by PE and P routers to 511
exchange routing information to facilitate dynamic LSP signaling: 512
• OSPF (RFC 2328 [9]) 513
• IS-IS (RFC 1195 [7]) 514
515
[R-9] Traffic engineering extensions of OSPF and IS-IS are used to exchange traffic 516
attributes for RSVP-TE tunnels. If TE is supported, both of the following methods MUST 517
be supported by PE and P routers: 518
• OSPF-TE (RFC 3630 [16]) 519
• IS-IS-TE (RFC 5305 [30]) 520
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 27 of 65
8 OAM 521
OAM in Carrier Ethernet networks was developed to provide fault management and performance 522
monitoring tools for network links and end-to-end EVCs. Figure 2 below shows the components 523
of OAM when the MEF services are provided using EVPN. 524
525
526 527
Figure 2 Components of OAM 528
529
Ethernet OAM 530
The OAM functions for the UNI should use the Ethernet OAM as defined in Link OAM (Clause 531
57/IEEE 802.3 [5]) and/or ITU-T G.8013/Y.1731 [40]. 532
8.1.1 Link OAM 533
The PE supports Ethernet Link OAM, when the user is directly connected to the network 534
demarcation point (i.e., at the MEF UNI in Figure 2). Link OAM provides OAM functions for 535
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 28 of 65
network access segments (UNI-C to UNI-N). Link OAM provides for Ethernet Link Fault 536
Detection, Monitoring and Loopback for access links. 537
538
[R-10] The PE MUST support link OAM Active mode as per Clause 57.2.9.1/IEEE 802.3 539
[5]. 540
541
[R-11] The PE MUST support initiating OAM Discovery process as per Subclause 542
57.3.2.1/IEEE 802.3 [5]. 543
544
[R-12] The PE MUST support sending informational OAM Protocol Data Units 545
(OAMPDU) as per Subclause 57.2.10/IEEE 802.3 [5]. 546
547
[R-13] The PE MUST support sending Event Notification OAMPDUs as per Subclause 548
57.2.10/IEEE 802.3 [5]. 549
550
[R-14] The PE MUST support sending loopback control OAMPDUs as per Subclause 551
5.2.11/IEEE 802.3 [5]. 552
553
[R-15] The PE MAY support sending Organization specific OAMPDU per Subclause 554
57/IEEE 802.3 [5]. 555
556
[R-16] The PE MAY support sending Variable Request OAMPDUs as per Subclause 557
57.4.3.3/IEEE 802.3 [5]. 558
8.1.2 ITU-T G.8013/Y.1731 559
The OAM functions defined in ITU-T G.8013/Y.1731 can be used for OAM of the UNI between 560
the CE and the PE. 561
562
[R-17] The PE MUST support sending and receiving OAM frames at level 0 (as 563
recommended in G.8013/Y.1731 [40]). 564
MEF Service OAM 565
The Carrier Ethernet Services are provided between one User Network Interface (UNI) and one or 566
more UNIs. A network operator must be able to manage the services using Service OAM 567
(SOAM). The network operator’s service OAM is originated at the PE’s UNI-N. 568
569
[R-18] The PE MUST support SOAM at the EVC SOAM level 4 as described in MEF 30 570
[43]. 571
572
[R-19] OAM frames, sent at SOAM levels 5, 6, or 7, as described in MEF 30, are sent as 573
user data and MUST be carried transparently. 574
575
[R-20] The PE MAY support sending and receiving SOAM frames across the UNI at the 576
UNI SOAM level 1, as described in MEF 30 [43]. 577
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 29 of 65
See Section 12.4.8 for information on performance monitoring. 578
MPLS OAM 579
This section describes techniques to perform OAM for the underlying MPLS tunnels used to 580
support Ethernet services. OAM is an important and fundamental functionality in an MPLS 581
network. OAM contributes to the reduction of operational complexity, by allowing for efficient 582
and automatic detection, localization, handling and diagnosis of defects. OAM functions, in 583
general, are used for fault-management, performance monitoring, and protection-switching 584
applications. 585
8.3.1 LSP OAM 586
This section describes techniques to perform OAM for the underlying MPLS LSPs used in an 587
EVPN application. 588
589
LSP-Ping and Bidirectional Forwarding Detection (BFD) RFC 5880 [32] are OAM mechanisms 590
for MPLS LSPs RFC 5884 [34]. Further it is desirable that the OAM traffic is sent in-band in an 591
LSP. The following OAM mechanisms are supported: 592
593
[R-21] The PE MAY support GAL and G-ACh per LSP, as per RFC 5586 [31]. 594
BFD for MPLS LSPs 595
BFD monitors the integrity of the LSP for any loss of continuity defect. In particular, it can be 596
used to detect a data plane failure in the forwarding path of an MPLS LSP. 597
598
[R-22] PE and P routers MUST support BFD for MPLS LSPs as per RFC 599
5884 [34]. 600
Detecting MPLS Data Plane Failures 601
LSP Ping is used to perform on-demand Connectivity Verification (CV) and Route Tracing 602
functions. It provides two modes: “ping” mode and “traceroute” mode. 603
In "ping" mode (basic connectivity check), the packet should reach the end of the path, at which 604
point it is sent to the control plane of the egress LSR, which then verifies whether it is indeed an 605
egress for the FEC. 606
607
[R-23] PE and P routers MUST support “ping” mode as per RFC 4379 [22]. 608
609
RFC 6424 [35] enhances the mechanism for performing Label Switched Path Ping (LSP Ping) 610
over MPLS Tunnels and when LSP stitching [RFC5150] is in use. 611
612
[R-24] PE and P routers MUST support enhanced MPLS ping and traceroute as per RFC 613
6424 [35]. 614
615
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 30 of 65
In "traceroute" mode (fault isolation), the packet is sent to the control plane of each transit LSR, 616
which performs various checks that it is indeed a transit LSR for this path; this LSR also returns 617
further information that helps check the control plane against the data plane. 618
619
[R-25] PE and P routers SHOULD support “traceroute” mode as per RFC 4379 [22]. 620
621
The LSP Ping Reply modes as defined in Section 3/RFC 4379 [22] apply as shown in 622
Table 1. 623
624
625
Reply Mode Echo request Echo Reply
Reply via an IPv4/IPv6 UDP packet (code value 2) MUST MUST
Reply via application level control channel (code value 4) MAY MAY
626
Table 1 LSP Ping Reply Modes 627
628
The following subsections of Section 3.2/RFC 4379 [22] concerning Target FEC Stack apply as 629
follows: 630
631
[R-26] When LDP is supported - LDP IPv4 prefix as defined in Section 3.2.1/RFC 4379 632
[22] MUST be supported. 633
634
[R-27] When RSVP is supported - RSVP IPv4 LSP as defined in Section 3.2.3/RFC 4379 635
[22] MUST be supported. 636
637
[R-28] When BGP is supported – BGP-labeled IPv4 prefix as defined in Section 638
3.2.11/RFC 4379 [22] MUST be supported. 639
640
[R-29] When LDP is supported - LDP IPv6 prefix as defined in Section 3.2.2/RFC 4379 641
[22] SHOULD be supported. 642
643
[R-30] When RSVP is supported - RSVP IPv6 LSP as defined in Section 3.2.4/RFC 4379 644
[22] SHOULD be supported. 645
646
[R-31] When BGP is supported – BGP-labeled IPv6 prefix as defined in Section 647
3.2.12/RFC 4379 [22] SHOULD be supported. 648
8.3.2 Convergence 649
This section specifies the requirements for the recovery mechanisms from PE to CE network (AC 650
link) failures. The recovery procedures are described in Section 17/RFC 7432 [39]. 651
652
[R-32] The PE routers MUST support recovery from PE to CE network failures (AC link 653
failures) as per Section 17.3/RFC 7432 [39]. 654
655
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 31 of 65
[R-33] The PE routers MUST support recovery from PE failures as per Section 17.2/RFC 656
7432 [39]. 657
658
659
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 32 of 65
9 QoS 660
The MPLS network supporting the Carrier Ethernet services has to provide QoS and service level 661
agreements. The QoS capabilities must be end-to-end, which includes both ACs and MPLS 662
domains. Usually an MPLS network will support guaranteeing sufficient bandwidth if available to 663
support new and existing Carrier Ethernet connections conforming to all SLA metrics including 664
protection mechanisms. 665
666
DiffServ-TE classes of service is used to support MEF 23.1 [45] “CoS Labels” to achieve a 667
particular level of performance. MPLS DiffServ-TE enables the advantages of both DiffServ and 668
TE. The DiffServ-TE requirement is to make separate bandwidth reservations for different classes 669
of traffic. RFC 3564 [14] provides the concept of a class type (CT). 670
671
The following capabilities are to be supported by the PEs: 672
673
[R-34] The PE MUST support at least 4 CoS and associated service metrics (e.g. delay, 674
delay variation, packet loss) as defined in MEF 22.1 [44] “EVC Requirements”. 675
676
[R-35] The PE SHOULD support Connection Admission Control to guarantee sufficient 677
bandwidth is available to support new connection conforming to all SLA metrics defined in 678
MEF 10.2 [42]. 679
680
[R-36] The PE SHOULD support Differentiated Service aware MPLS traffic engineering 681
as per RFC 4124 [19]. 682
683
[R-37] The ingress PE MUST map the PCP (in the PRI field of the 802.1Q VLAN tag 684
IEEE 802.1Q [6]) into the TC field of the MPLS label stack. 685
Tunnel CoS mapping and marking 686
Two types of LSPs are defined in RFC 3270 [11]: 687
688
[R-38] The PE and P routers MUST support E-LSP as per Section 1.2/RFC 3270 [11]: 689
LSPs which can transport multiple Ordered Aggregates, so that the TC field of the MPLS 690
Shim Header conveys to the LSR the PHB to be applied to the packet (covering both 691
information about the packet's scheduling treatment and its drop precedence). 692
693
[R-39] The PE and P routers MAY support L-LSP as per Section 1.3/RFC 3270 [11]: LSPs 694
which only transport a single Ordered Aggregate, so that the packet's scheduling treatment 695
is inferred by the LSR exclusively from the packet's label value while the packet's drop 696
precedence is conveyed in the TC field of the MPLS Shim Header. 697
698
[R-40] The PE MUST support COS marking in the TC bits of the LSP labels. 699
700
[R-41] The PE MUST support the Pipe model as per RFC 3270 [11]. 701
702
703
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 33 of 65
10 PSN resiliency 704
In EVPN, the PEs are connected over an underlying PSN infrastructure. For EVPN resiliency, the 705
PSN must necessarily be resilient. When the PEs are connected by an MPLS infrastructure then 706
the resiliency mechanisms in MPLS such as fast reroute (FRR) are required. This section lists the 707
resiliency requirements for MPLS. If the MPLS infrastructure is run over another layer (e.g. a L1 708
network) the resiliency requirements of the other layers are considered to be outside the scope of 709
this section. When resiliency mechanisms are available at multiple layers the resiliency 710
mechanism at a layer must be triggered only after a sufficient delay to let the resiliency mechanism 711
of the underlying layer to take effect. 712
713
MPLS resiliency requires failure detection mechanisms and LSP recovery mechanisms. To speed 714
up total recovery time, local-repair mechanisms with pre-computed, pre-established 715
alternate/backup paths should be used whenever possible. Section 10.1 lists the failure detection 716
requirements and Section 10.2 lists the requirements for LSP recovery. 717
718
MPLS resiliency is also affected by the restart of control-plane protocols. The MPLS requirements 719
to support resiliency of control protocols are listed in Section 10.3. 720
Failure detection 721
The failure detection mechanism that triggers the recovery mechanisms should have a low failure 722
detection time and also a low overhead. In order for the deployment to allow a choice of routing 723
protocols, the failure detection mechanism should be independent of specific routing protocols. 724
The Bidirectional Forwarding Detection (BFD) protocol specified in RFC 5880 [32] provides such 725
a mechanism. For MPLS LSP BFD requirements see Section 8.3.1. 726
727
[R-42] The PE and P routers MUST support BFD for single hops as per RFC 5881 [33] 728
LSP recovery 729
The LSP recovery mechanism should support local repair mechanisms with pre-computed and pre-730
established alternate/backup paths for both RSVP-TE RFC 3209 [10] and LDP RFC 5036 [25] 731
signaled LSPs. Recovery from different types of failure such as link, node, etc. should be 732
supported. 733
734
[R-43] The PE and P routers MUST support the facility backup method of doing fast 735
reroute (FRR) for RSVP-TE LSP Tunnels as per RFC 4090 [18]. 736
737
[R-44] The PE and P routers SHOULD support the one-to-one backup method of doing 738
fast reroute (FRR) for RSVP-TE LSP Tunnels as per RFC 4090 [18]. 739
740
[R-45] The PE and P routers MUST support the loop-free alternates (LFA) method of FRR 741
for LDP LSPs as per RFC 5286 [29] as well as support LFA FRR for the IGP on whose 742
routes LDP depends. 743
744
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 34 of 65
Control plane resiliency 745
To prevent LSPs from going down due to control-plane protocols restart, the graceful restart 746
control-plane resiliency mechanism is required. 747
748
[R-46] The PE and P routers MUST support RSVP-TE graceful restart as specified in 749
Section 9/RFC 3473 [12] as well as graceful restart for the routing protocols on which 750
RSVP-TE path computation depends. 751
752
[R-47] The PE and P routers MUST support LDP graceful restart as specified in RFC 3478 753
[13] as well as graceful restart for the routing protocols on whose routes LDP depends. 754
755
[R-48] The PE and P routers SHOULD support OSPF graceful restart as specified in RFC 756
3623 [15]. 757
758
[R-49] The PE and P routers SHOULD support IS-IS graceful restart as specified in RFC 759
5306 [17]. 760
761
762
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 35 of 65
11 BGP MPLS-Based Ethernet VPN 763
This section covers the generic BGP MPLS-based Ethernet VPN requirements. Specific 764
requirements such as multicast that are applicable to a subset of Ethernet VPN services (e.g. EP-765
LAN, EVP-LAN, etc) are covered in subsequent sections. 766
767
EVPN overcomes the limitations of current E-Line and E-LAN services supported by VPLS (RFC 768
4761 [23], RFC 4762[24]) and VPWS. EVPN provides flexible multi-homing with all-active 769
redundancy mode, MAC learning using control plane, multicast optimization, provisioning 770
simplicity and network resiliency between edge nodes. 771
772
The EVPN specification supports several ways for PE nodes to connect, but this TR only supports 773
use of MPLS, which enable easy interworking with TR-224 [4] based Ethernet services. 774
Reference Architecture and Overview 775
Figure 3 describes EVPN for the next-generation of Ethernet services. An EVPN instance 776
comprises of CEs connected to PEs, which are part of the MPLS network. The PEs provide virtual 777
Layer 2 bridge connectivity between CEs. The PEs are connected by an underlying MPLS 778
network that provides QoS and resiliency. 779
780
Unlike VPLS, which uses only data-plane based MAC learning, EVPN uses the control plane 781
based MAC learning for remote MACs. EVPN uses MP-BGP to distribute MAC routes and 782
allows fine-grained control over MAC route distribution. 783
784
EVPN instance (EVI) is an EVPN routing and forwarding instance on a PE. If a CE is multi-785
homed to two or more PEs, the set of Ethernet links constitute an Ethernet Segment (ES). Each 786
Ethernet Segment is identified using a unique Ethernet Segment identifier (ESI). For additional 787
details see RFC 7432 [39]. 788
789
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 36 of 65
790
791 Figure 3 EVPN Architecture for Ethernet services using BGP MPLS 792
793
EVPN Service Interfaces 794
EVPN defines several types of service interfaces. See Section 6/RFC 7432 [39] for details. These 795
service interfaces are consistent with MEF-defined services and provide easy migration to EVPN 796
infrastructure for even richer service offerings. The various types of service interfaces include 797
mapping of specific VLANs or their bundles and even allow service awareness when they are 798
mapped to EVPN instances. The requirements for the support of various service interfaces are 799
specified in the subsections of the respective services. 800
11.2.1 VLAN-Based Service Interfaces 801
This service interface supports a single broadcast domain or VLAN per-EVPN instance. 802
This service interface can be used to support E-LAN or E-Line for a single broadcast domain with 803
customer VLANs having local significance. Ethernet frames transported over an MPLS network 804
remain tagged with the originating VID. VID translation can be performed on the destination PE. 805
806
11.2.2 VLAN Bundle Service Interfaces 807
This service interface supports a bundle of VLAN over one EVPN instance. Multiple VLANs 808
share the same bridge. It supports an N:1 mapping between VLAN ID and MAC-VRF. This 809
service interface requires that MAC addresses are unique across VLANs of the EVI and VID 810
translation is not allowed. 811
812
This service interface also supports a special case known as a port-based VLAN Bundle service 813
interface, where all the VLANs on a port are part of the same service and map to the same bundle. 814
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 37 of 65
11.2.3 VLAN-Aware Bundle Service Interfaces 815
This service interface is an additional service interface defined in EVPN that is not supported by 816
TR-224 or VPLS. It provides customers with a single E-LAN service for multiple broadcast 817
domains. With this service interface, an EVPN instance consists of multiple broadcast domains or 818
VLANs, with each VLAN having its own bridge domain. Like the VLAN bundle service 819
interface, this interface supports N:1 mapping between VLAN ID and EVI. Since bridge domains 820
are separate, it allows for local VID translation. 821
822
This service interface also supports a special case known as a port-based VLAN-Aware bundle 823
service interface, where all the VLANs on a port are part of the same service and map to the same 824
bundle. 825
Data Plane 826
11.3.1 Underlying PSN transport 827
[R-50] The PEs MUST support MPLS as the underlying PSN transport as specified in 828
Section 4/RFC 7432 [39]. 829
11.3.2 VPN encapsulation 830
To distinguish packets received over the PSN destined to different EVPN instances, MPLS labels 831
must be used as described in Section 4/RFC 7432 [39]. The specific data plane operations 832
applicable to a service are specified in the subsections of the respective service. 833
11.3.3 VID Translation 834
[R-51] The PEs MUST support VID translation for packets received from the PSN and sent 835
to the CE, when supporting service interfaces as specified in Section 6/RFC 7432 [39]. 836
11.3.4 Frame Ordering 837
Section 18/RFC 7432 specifies frame ordering. In order to avoid misordering, it is recommended 838
that P routers not use deep packet inspection to do ECMP. 839
840
[R-52] The P routers SHOULD NOT do deep packet inspection for ECMP. RFC 6790 [36] 841
specifies techniques so that P routers do effective load balancing without the need for deep 842
packet inspection. 843
844
Control Plane 845
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 38 of 65
The EVPN PEs signal and learn MAC address over the control plane. RFC 7432 adds BGP 846
extended communities, which allow PE routers to advertise and learn MAC addresses and Ethernet 847
segments. This is one of the major differences with the VPLS solution, which relies on data-plane 848
learning. EVPN added four Route types and communities. For additional details see RFC 7432 849
[39]. 850
851
[R-53] The PEs MUST support MP-BGP as a control protocol for EVPN as specified in 852
Section 4 and 7/RFC 7432 [39]. 853
854
Note: The detailed control protocol requirements of MP-BGP are specified in the 855
subsections of the respective services. 856
857
With MPLS data plane, BGP routes also signal the MPLS labels associated with MAC addresses 858
and Ethernet segments. This separates EVPN from a VPLS solution. EVPNs do not use 859
Pseudowires. 860
Multi-Homing and Load balancing 861
Due to rapid increase of data traffic, running the network in active/standby mode can be 862
inefficient. In addition to better link utilization, multi-homed connections also offer greater 863
resiliency and reliability against the failure of one connection or node. Multi-homing includes the 864
ability of establishing multiple connections between PEs and to load-balance across those 865
connections. For additional details on multi-homing see Section 8/RFC 7432 [39]. EVPNs 866
supports both single-active and all-active multi-homing with load balancing. VPLS only supports 867
single-active multi-homing. 868
869
With support for both all-active per-service and all-active per-flow multi-homing, EVPNs enables 870
better load balancing across peering PEs as compared to VPLS that cannot load balance across 871
peering PEs. 872
873
It must be possible to connect a CE to two or more PEs for purposes of multi-homing and load 874
balancing as specified in Section 8 and 14/RFC 7432 [39]. 875
11.5.1 All-Active Redundancy Mode 876
All-active redundancy mode allows the CE device to connect via a “single” Ethernet bundle to 877
multiple PEs using LAG. All the PEs must be allowed to forward traffic to/from that Ethernet 878
Segment. 879
880
[R-54] PE router MUST support “All-Active redundancy mode” as specified in Section 881
14/RFC 7432 [39]. 882
883
11.5.2 Single-Active Redundancy Mode 884
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 39 of 65
In this mode, when a CE is connected to two or more PEs over an Ethernet segment, only a single 885
PE must be allowed to forward traffic to/from that Ethernet Segment. In this mode the CE device 886
connect via “separate” Ethernet bundles to multiple PEs. 887
888
[R-55] PE router MUST support “Single-Active redundancy mode” as specified in Section 889
14/RFC 7432 [39]. 890
Fast Convergence 891
Section 17/RFC 7432 provides failure recovery from different types of network failures. VPLS 892
relies on the underlying MPLS capabilities such as Fast Reroute. Lack of all-active multi-homing 893
in VPLS makes it difficult to achieve fast restoration in case of an edge node or edge link failure. 894
895
[R-56] The PEs MUST support convergence as specified in Section 17/RFC 7432 [39]. 896
897
898
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 40 of 65
12 EVPN enabled multipoint to multipoint Ethernet VPN services (E-LAN) 899
The EVPN technology enables the creation of multipoint-to-multipoint Ethernet VPN services 900
over an MPLS network. EVPN can be used to create the EP-LAN and EVP-LAN services of the 901
E-LAN service type defined by MEF 6.2. A high-level reference architecture of how these 902
services are architected using EVPN along with the list of the supported service attributes is 903
described in Section 12.1 and 12.2. In addition to the Carrier Ethernet defined service 904
characteristics, EVPN significantly enhances important service characteristics such as reliability 905
and scalability. The EVPN requirements for multipoint-to-multipoint Ethernet VPN services are 906
listed in Section 12.3. 907
Ethernet Private LAN (EP-LAN) 908
The Ethernet Private LAN (EP-LAN) service uses a multipoint-to-multipoint EVC. In a 909
multipoint EVC, two or more UNIs must be associated with one another. The EP-LAN service is 910
defined to provide CE-VLAN tag preservation and tunneling of key Layer 2 control protocols. A 911
key advantage of this service is that VLANs can be configured across the sites without any need to 912
coordinate with the service provider. 913
914
EP-LAN provides connectivity to customers with multiple sites, such that all sites appear to be on 915
the same local area network. Each interface is configured for “All to One Bundling”. EP-LAN 916
supports CE-VLAN CoS preservation. Service multiplexing is disabled on the UNI. 917
918
919
920 Figure 4 Ethernet Private LAN (EP-LAN) Service 921
922
923
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 41 of 65
Ethernet Virtual Private LAN (EVP-LAN) 924
The Ethernet Virtual Private LAN (EVP-LAN) service allows service multiplexing at the UNI. It 925
allows users of an E-LAN service type to interconnect their UNIs and at the same time access 926
other services (e.g. E-Line). Figure 5 shows an example of multiple services access from a single 927
UNI. In this example, the user has an EVP-LAN service for multipoint data connectivity and an 928
EVPL service (P2P EVC) for accessing a value-add service from one of the UNIs. 929
930
Bundling can be used on the UNI in the EVP-LAN service and supports CE-VLAN tag 931
preservation. All to One Bundling is disabled. 932
933
934 Figure 5 Ethernet Virtual Private LAN (EVP-LAN) Service 935
936
EVPN for establishing EP-LAN and EVP-LAN 937
Section 5 provides an overview of PPVPN in MPLS networks. It also outlines the comparison of 938
Layer 2 Ethernet VPNs in MPLS networks using VPLS and EVPNs. 939
940
EVPN provides support for the E-LAN service type in MPLS networks as described in Section 11. 941
RFC 7432 describes procedures for BGP MPLS-based Ethernet VPNS. EVPN requires extensions 942
to existing IP/MPLS protocols. EVPN supports both provisioning and signaling for Ethernet 943
VPNs and incorporate flexibility for service delivery over Layer 3 networks. 944
945
The PE supports BGP MPLS-based Ethernet VPN signaling and provisioning as specified in [R-946
53] of Section 11.4. 947
948
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 42 of 65
12.3.1 Service Interfaces 949
EVPN supports several service connectivity options for delivering MEF services. They are 950
provided in Section 11.2. MEF service requirements for EP-LAN and EVP-LAN are different. 951
For EP-LAN, each interface is configured for “All to One Bundling”. For EVP-LAN “All to One 952
Bundling” is disabled. VLAN-Aware Bundle service interface is not supported in VPLS-based 953
implementation (TR-224). When interworking with TR-224, it is recommended not to use VLAN-954
Aware Bundle service interface. This service interface provides a more flexible service offering 955
and is commonly used for data center interconnect. 956
Service Interfaces for EP-LAN 957
[R-57] The PE routers MUST support Port-based Service Interface as defined in Section 958
6.2.1/RFC 7432 [39]. 959
960
[R-58] The PE routers SHOULD support Port-Based VLAN-Aware Service Interface as 961
defined in Section 6.3.1/RFC 7432 [39]. 962
12.3.1.1.1 Data Center Considerations 963
If the PE router is designed for use in a data center interconnect environment, the following 964
requirement is applicable: 965
966
[R-59] The PE routers MUST support Port-Based VLAN-Aware Service Interface as 967
defined in Section 6.3.1/RFC 7432 [39]. 968
Service Interfaces for EVP-LAN 969
[R-60] The PE routers MUST Support VLAN-based Service Interface as defined in Section 970
6.1/RFC 7432 [39]. 971
972
[R-61] The PE routers MUST support VLAN Bundle Service Interface as defined in 973
Section 6.2/RFC 7432 [39] 974
975
[R-62] The PE routers SHOULD support VLAN-Aware Bundle Service Interface as 976
defined in Section 6.3/RFC 7432 [39]. 977
12.3.1.2.1 Data Center Considerations 978
If the PE router is designed for use in a data center interconnect environment, the following 979
requirement is applicable: 980
981
[R-63] The PE routers MUST support VLAN-Aware Bundle Service Interface as defined 982
in Section 6.3/RFC 7432 [39]. 983
984
985
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 43 of 65
12.3.2 Data plane 986
The requirements for data plane per Section 11.3 are applicable. 987
Local learning 988
[R-64] The PE MUST be able to do data-plane learning of MAC addresses using IEEE 989
Ethernet learning procedures for packets received from the CEs connected to it as specified 990
in Section 9.1/RFC 7432 [39]. 991
Remote learning 992
[R-65] The PE MUST be able to do control-plane learning of MAC addresses using MP-993
BGP’s MAC Advertisement route for CEs that are connected to remote PEs as specified in 994
Section 9.2/RFC 7432 [39]. 995
12.3.3 Tunnel signaling 996
The PEs are connected by MPLS Label Switch Paths (LSPs) acting as PSN tunnels. Traffic 997
Engineered PSN tunnels must be used when specific path (e.g. for protection purpose), QoS, or 998
bandwidth constraints are required. 999
1000
[R-66] PE and P routers MUST support dynamic signaling to setup both TE LSPs and 1001
routed LSPs. See Section 7.1 for details. 1002
12.3.4 Routing 1003
The requirements for routing per Section 7.2 are applicable. 1004
12.3.5 Multi-Homing and Load balancing 1005
The requirements for multi-homing and load balancing per Section 11.5 are applicable. 1006
Load balancing at intermediate nodes 1007
When load balancing, packets that belong to a given ‘flow’ must be mapped to the same port. 1008
Intermediate P nodes have no information about the type of the payload inside the LSP. 1009
Intermediate LSR should make a forwarding choice based on the MPLS label stack. In order to 1010
avoid any mis-ordering of frames, the requirements specified in Section 11.3.4 apply. 1011
1012
The PE that has knowledge of the Ethernet service (e.g. Bundling or multiclass service) can take 1013
further action. IETF RFC 6790 [36] provide methods of assigning labels to flows, or flow groups, 1014
such that Label Switching Routers can achieve better load balancing. 1015
1016
[R-67] The PE SHOULD support Entropy Labels as per RFC 6790 [36]. 1017
1018
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 44 of 65
1019
1020
1021
12.3.6 OAM 1022
Ethernet Link OAM 1023
The Ethernet link OAM is supported as per Section 8.1.1. 1024
Label Switched Paths (LSP) OAM 1025
LSP OAM is supported as per Section 8.3.1. 1026
MEF Service OAM 1027
MEF service is supported as per Section 8.1.2. 1028
12.3.7 Convergence 1029
Failure recovery from different types of network failure is supported as per Section 8.3.2. 1030
12.3.8 PSN Resiliency 1031
PSN resiliency is supported as per Section 10. 1032
1033
Fast Convergence 1034
Fast convergence is supported as per Section 11.6. 1035
12.3.9 Multicast and Broadcast 1036
[R-68] PE routers SHOULD support multicast and broadcast traffic as per Section 16/RFC 1037
7432 [39]. 1038
12.3.10 QoS 1039
In general, an E-LAN service type can provide a best effort service with no performance 1040
assurance. In certain cases, an E-LAN service type can be defined with performance objectives 1041
(see Section 9.2/MEF 6.2 [50]. 1042
1043
[R-69] PE routers SHOULD support the QoS mapping as per Section 9. 1044
12.3.11 Security 1045
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 45 of 65
[R-70] PE routers MUST support security as per Section 19/RFC 7432 [39]. 1046
Support of service attributes for EP-LAN and EVP-LAN 1047
Section 9.2/MEF 6.2 [50] specifies the E-LAN service type that is the basis for LAN services. 1048
Section 10.3 and 10.4/MEF 6.2 [50] provides service attributes and parameters for EP-LAN and 1049
EVP-LAN services respectively. TR-224 refers to MEF 6.1 and MEF 10.2. TR-350 uses the 1050
backward compatible subset of the revised specifications MEF 6.2 [50] (see Appendix B) and 1051
MEF 10.3 [49] to achieve the equivalent function. 1052
1053
Some of the service attributes and parameters are provided by Ethernet physical interface and 1054
service provisioning (e.g., Physical medium, Speed, Mode, MAC layer, EVC type, maximum 1055
number of EVCs, etc.). This section only describes those service attributes and parameters that are 1056
relevant to transporting the EVPN traffic over PSN. 1057
12.4.1 Bandwidth Profile 1058
A bandwidth profile defines how rate enforcement of Ethernet frames is applied at an UNI. 1059
Bandwidth profiles enable offering service bandwidth below the UNI access speed (aka Speed) 1060
and limit the amount of traffic entering the network per the terms of the SLA. 1061
1062
For LAN services, bandwidth profiles can be optionally specified per-UNI (ingress and egress), 1063
per-EVC (ingress and egress), and/or per CoS (ingress and egress). An E-LAN service can be 1064
provided as a best effort service without any bandwidth guarantee. 1065
1066
[R-71] A PE SHOULD support the bandwidth profile algorithm as per the portion of 1067
Section 12/MEF 10.3 [49] that is backward compatible with MEF 10.2 [42]. 1068
1069
In order to support bandwidth profile, technique such as admission control and Diffserv-TE as 1070
specified in Section 9 are used. 1071
12.4.2 Bundling 1072
Section 9.12/MEF 10.3 [49] specifies the bundling service attribute. Bundling implies “A UNI 1073
attribute in which more than one CE-VLAN ID can be associated with an EVC”. All to one 1074
bundling enabled is a special case of bundling. It implies “A UNI attribute in which all CE-VLAN 1075
IDs are associated with a single EVC”. Table 12/MEF 10.3 [49] provides valid combinations for 1076
“All to one bundling” and “Service multiplexing” attributes. 1077
1078
EP-LAN must have “All to one bundling” attribute enabled. For EVP-LAN the bundling attribute 1079
can be enabled or disabled. However, for EVP-LAN “All to one bundling” must be disabled. 1080
1081
12.4.3 CE-VLAN ID preservation for EVC 1082
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 46 of 65
CE-VLAN ID preservation service attribute defines whether the CE-VLAN ID is preserved 1083
(unmodified) across the EVC. 1084
1085
For EP-LAN, CE-VLAN ID preservation must be enabled and CE-VLAN ID is preserved for EVC 1086
over the PSN. 1087
1088
For EVP-LAN, if CE-VLAN ID preservation is enabled, CE-VLAN ID is preserved for EVC over 1089
the PSN. 1090
1091
Note: If CE-VLAN ID preservation is enabled, No VID translation is supported for the EVC. 1092
EVP-LAN can support VID translation when using EVPN service type “VLAN-Based Service 1093
type” and “VLAN-Aware Bundle service interface”. 1094
12.4.4 CE-VLAN CoS preservation for EVC 1095
CE-VLAN CoS preservation service attribute defines whether the CE-VLAN CoS bits are 1096
preserved (unmodified) across the EVC. 1097
1098
For EP-LAN, CE-VLAN CoS preservation must be enabled (see Table 15/MEF 6.2 [50]) and CE-1099
VLAN CoS is preserved for EVC over PSN. 1100
1101
For EVP-LAN, CE-VLAN CoS preservation can be either enabled or disabled (see Table 18/MEF 1102
6.2). In an EVC with CE-VLAN CoS preservation is enabled, the EVPN preserves the CoS bits 1103
over PSN. 1104
12.4.5 EVC Maximum Service Frame Size 1105
The mapping from “EVC Maximum Service Frame Size” to “EVC MTU” is provided in Table 40 1106
Appendix B of MEF 6.2 [50]. 1107
1108
The EVC Maximum Service Frame Size size is configurable with a default value of 1600 byte. 1109
1110
When Ethernet frames are transported in MPLS networks, the MPLS packet includes the labels, 1111
and the EVC frame as payload. The path MTU is the largest packet size that can traverse this path 1112
without fragmentation. The ingress PE can use Path MTU Discovery to find the actual path MTU. 1113
1114
[R-72] PE SHOULD support configurable EVC Maximum Service Frame Size of at least 1115
1600 bytes (see Table 6/MEF 6.2). 1116
1117
1118
1119
1120
12.4.6 Frame delivery 1121
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 47 of 65
The frame delivery policy rules enable the service provider to specify how different frame types 1122
are handled by a PE. They enable setting specific rules for forwarding, discarding or conditionally 1123
forwarding specific frame types. The frame types used by the rules are: 1124
1125
• Unicast 1126
• Multicast 1127
• Broadcast 1128
1129
[R-73] PE MUST support setting policy function of frame delivery rules for 1130
forwarding, discarding or conditionally forwarding unicast, multicast and 1131
broadcast frames per EP-LAN and EVP-LAN services. 1132
12.4.7 Layer 2 control protocols 1133
The Layer 2 control protocol processing is independent of the EVC at the UNI. L2CP handling 1134
rules are set according to the definition of Section 8/MEF 6.1.1 [48] and differ per-service type. 1135
The PE policy function supports setting of rules for handling L2CP per service type. 1136
1137
EVC L2CP handling per service type can be set to: 1138
• Discard – Drop the frame. 1139
• Peer can be applicable: For example L2CP/LAMP, Link OAM, Port Authentication, and E-1140
LMI. 1141
• Tunnel – Pass to the egress UNI. 1142
1143
[R-74] PE MUST support policy function setting of rules for handing L2CP per service 1144
type as specified in Section 8/MEF 6.1.1 [48]. 1145
1146
Note: This specification only supports MEF 6.1.1 for L2CP processing 1147
requirements. Support of multiple-CEN L2CP MEF 45 [51] is outside the scope of 1148
this document. 1149
12.4.8 EVC performance 1150
The performance parameters indicate the quality of service for that service instance. They consist 1151
of the following: 1152
• Availability 1153
• Frame Delay 1154
• Frame delay variation 1155
• Frame loss ratio 1156
1157
The requirements for support of CoS and mapping are specified in QoS Section 9. 1158
1159
[R-75] The PE MUST support MEF SOAM performance monitoring as per MEF 35 [47]. 1160
1161
For transport of SOAM see Section 8.2. 1162
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 48 of 65
1163
1164
1165
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 49 of 65
13 EVPN enabled point to point Ethernet VPN services (E-Line) 1166
The Ethernet Line service is defined in MEF 6.2 [50] and is based on a point-to-point Ethernet 1167
Virtual Connection (EVC). Virtual Private Wire Service (VPWS) is a Layer 2 VPN service used 1168
to emulate the E-Line service type in an MPLS network. draft-ietf-bess-evpn-vpws Error! R1169
eference source not found. describes how EVPN can be used to support VPWS in IP/MPLS 1170
networks. The use of EVPN mechanisms for VPWS brings the benefits of EVPN to point-to-point 1171
services. EVPN can be used to create the Ethernet Private Line (EPL) and Ethernet Virtual Private 1172
Line (EVPL) services of the E-Line service type defined by MEF 6.2 [50]. 1173
1174
A high-level reference architecture of how these services are architected using EVPN along with 1175
the list of the supported service attributes is described in Section 13.1 and 13.2. In addition to the 1176
Carrier Ethernet-defined service characteristics, EVPN significantly enhances important service 1177
characteristics such as reliability and scalability. The EVPN-VPWS requirements for point-to-1178
point Ethernet VPN services are listed in Section 13.3. 1179
Ethernet Private Line (EPL) 1180
Ethernet Private Line (EPL), uses a point-to-point EVC between two UNIs to provide a high 1181
degree of transparency for service frames between UNIs it interconnects. The service frames, 1182
headers, and most Layer 2 protocols are identical at both the source and destination UNI. It does 1183
not allow for service multiplexing; that is, a dedicated UNI (physical interface) is used for the 1184
EPL. The figure below shows EPL service. 1185
1186
A key advantage of this service is that VLANs can be configured across the sites without any need 1187
to coordinate with the service provider. Each interface is configured for “All to One Bundling”. 1188
EPL supports CE-VLAN CoS preservation. For cases where an ingress bandwidth profile is 1189
applied, the CE is expected to shape traffic. 1190
1191
1192
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 50 of 65
1193 1194
1195
1196
1197
Ethernet Private Line (EPL): 1198
• It replaces a TDM private line 1199
• Dedicated UNIs for Point-to-Point connections 1200
• Single Ethernet Virtual Connection (EVC) per UNI 1201
• The most popular Ethernet service type due to its simplicity 1202
Ethernet Virtual Private Line (EVPL) 1203
Ethernet Virtual Private Line (EVPL) uses a point-to-point EVC between two UNIs, but does not 1204
provide full transparency as with the EPL service. The EVPL service also allows for service 1205
multiplexing, which means that more than one EVC can be supported at the UNI. Because service 1206
multiplexing is permitted, some service frames may be sent to one EVC, while other service 1207
frames may be sent to other EVCs. The service definition for EVPL is specified in Section 1208
10.1/MEF 6.2 [50]. 1209
1210
EVPL is commonly used for connecting subscriber hub and branch locations. It allows users of an 1211
E-Line service type to interconnect their UNIs and at the same time access other services (e.g. E-1212
LAN). Figure 7 shows an example of multiple services access from a single UNI. In this example, 1213
the user has an EVP-LAN service for multipoint data connectivity and an EVPL service (P2P 1214
EVC) for accessing a value-add service from one of the UNIs. Bundling can be used on the UNI 1215
in the EVPL service and supports CE-VLAN tag preservation. All to One Bundling is disabled. 1216
Figure 6 Ethernet Private Line (EPL) Service
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 51 of 65
1217 1218
EVPN for establishing EPL and EVPL 1219
VPWS support in EVPN draft-ietf-bess-evpn-vpws Error! Reference source not found. describes h1220
ow EVPN can be used to support VPWS in IP/MPLS networks. EVPN enables the following 1221
service capabilities: single-active as well as all-active multi-homing; flow-based load balancing; 1222
and policy. It eliminates the need for PWs for point-to-point Ethernet services. EVPN has the 1223
ability to forward customer traffic at the PE without any MAC lookup because the MPLS label 1224
associated with the EVI is used. EVPL can be considered as a VPWS with only two-attachment 1225
circuits (AC) or MEF-UNIs. For additional details see draft-ietf-bess-evpn-vpws Error! R1226
eference source not found.. 1227
1228
RFC 7432 [39] describes procedures for BGP MPLS-based Ethernet VPNs. EVPN requires 1229
extensions to existing IP/MPLS protocols. EVPN supports both provisioning and signaling for 1230
Ethernet VPNs and incorporates flexibility for service delivery over Layer 3 networks. 1231
1232
1233
The PE supports BGP MPLS-based Ethernet VPN signaling and provisioning as specified in 1234
requirement [R-53] of Section 11.4. 1235
13.3.1 Service Interfaces 1236
EVPN-VPWS supports several service connectivity options for delivering MEF services. They are 1237
provided in Section 2/draft-ietf-bess-evpn-vpws Error! Reference source not found.. This s1238
ervice does not use “VLAN-Aware Bundle Service interface”. . 1239
1240
Figure 7 Ethernet Virtual Private Line (EVPL) Service
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 52 of 65
MEF service requirements for EPL and EVPL are different. For EPL, each interface is configured 1241
for “All to One Bundling”. For EVPL “All to One Bundling” is disabled. 1242
Service Interfaces for EPL 1243
[R-76] The PE routers MUST support Port-based Service Interface as defined in Section 1244
6.2.1/RFC 7432 [39]. 1245
1246
Service Interfaces for EVPL 1247
[R-77] The PE routers MUST Support VLAN-based Service Interface as defined in Section 1248
2.1/draft-ietf-bess-evpn-vpws Error! Reference source not found.. 1249
1250
[R-78] The PE routers MUST support VLAN Bundle Service Interface as defined in 1251
Section 2.2/draft-ietf-bess-evpn-vpws Error! Reference source not found.. 1252
1253
13.3.2 Operation 1254
PE routers must support point to point deployments as defined in Section 4/draft-ietf-bess-evpn-1255
vpws Error! Reference source not found.. 1256
13.3.3 Data Plane 1257
The requirements for data plane per Section 11.3.1 and 11.3.2 are applicable. 1258
Local learning 1259
The requirements for local learning per Section 12.3.2.1 are applicable. 1260
Remote learning 1261
The requirements for remote learning per Section 12.3.2.2 are applicable. 1262
13.3.4 BGP extensions 1263
EVPN-VPWS uses the Ethernet A-D per-EVI route to signal VPWS services. An EVPN instance 1264
MUST NOT be configured with both VPWS service instance and EVPN multipoint services. 1265
draft-ietf-bess-evpn-vpws Error! Reference source not found. Section 3.1 adds a EVPN Layer 2 a1266
ttributes extended community. 1267
1268
[R-79] The PE routers MUST support EVPN Layer 2 attributes extended community as 1269
defined in Section 3.1/draft-ietf-bess-evpn-vpws Error! Reference source not found.. 1270
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 53 of 65
13.3.5 Tunnel signaling 1271
The requirements for tunnel signaling per Section 12.3.3 are applicable. 1272
13.3.6 Routing 1273
The requirements for routing per Section 7.2 are applicable. 1274
13.3.7 Multi-Homing and Load balancing 1275
The requirements for multi-homing and load balancing per Section 11.5 are applicable. 1276
Load balancing at intermediate nodes 1277
The requirements for load balancing at intermediate nodes per Section 12.3.5.1 are 1278
applicable. 1279
13.3.8 OAM 1280
Ethernet Link OAM 1281
The Ethernet link OAM is supported as per Section 8.1.1. 1282
Label Switched Paths (LSP) OAM 1283
LSP OAM is supported as per Section 8.3.1. 1284
MEF Service OAM 1285
MEF service is supported as per Section 8.2. 1286
13.3.9 Convergence 1287
Failure recovery from different types of network failure is supported as per Section 8.3.2. 1288
13.3.10 PSN Resiliency 1289
PSN resiliency is supported as per Section 10. 1290
Fast Convergence 1291
Fast convergence is supported as per Section 11.6. 1292
13.3.11 QoS 1293
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 54 of 65
E-Line service type can be defined with performance objectives (see Section 9.1/MEF 6.2 [50]). 1294
PE routers MUST support the QoS mapping as per Section 9. 1295
13.3.12 Security 1296
[R-80] PE routers MUST support security as per Section 19/RFC 7432 [39]. 1297
Support of service attributes for EPL and EVPL 1298
Section 9.1/MEF 6.2 [50] specifies the Ethernet Line (E-Line) Service Type that is the basis for a 1299
broad range of Point-to-Point services. Section 10.1 and 10.2/MEF 6.2 [50] provides service 1300
definitions for the Ethernet Private Line (EPL) Service and the Ethernet Virtual Private Line 1301
(EVPL) Service that are based on the E-Line Service Type by defining the required Service 1302
Attributes and related parameter values. The service attributes are tabulated in Section 8/MEF 6.2 1303
[50]. 1304
1305
TR-224 refers to MEF 6.1 [41] and MEF 10.2 [42]. TR-350 uses the backward compatible subset 1306
of the revised specifications MEF 6.2 [50] and MEF 10.3 [49] to achieve the equivalent function. 1307
This addresses interworking with TR-224. 1308
1309
This section only describes those Service Attributes and parameters that are relevant to 1310
transporting the service over the PSN. 1311
13.4.1 Bandwidth Profile 1312
A bandwidth profile characterizes the lengths and arrival times of Service frames at a reference 1313
point. Bandwidth profiles enable offering service bandwidth below the UNI access speed (aka 1314
Speed) and limit the amount of traffic entering the network per the terms of the SLA. 1315
1316
For the EPL and EVPL service, bandwidth profiles can be optionally specified per-UNI (ingress 1317
and egress), per-EVC (ingress and egress), and/or per CoS (ingress and egress) as specified in 1318
Section 10.1 and 10.2/MEF 6.2 [50]. The EPL and EVPL service can be provided as a best effort 1319
service without any bandwidth guarantee. 1320
1321
Note: Bandwidth profiles can be optionally provisioned per UNI (ingress and egress). They are not 1322
carried by EVPN protocols. 1323
1324
[R-81] A PE SHOULD support the bandwidth profile algorithm as per the portion of 1325
Section 12/MEF 10.3 [49] that is backward compatible with MEF 10.2 [42]. 1326
1327
In order to support bandwidth profile, technique such as admission control and Diffserv-TE as 1328
specified in Section 9 are used. 1329
13.4.2 Bundling 1330
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 55 of 65
Bundling is defined in Section 2/MEF 10.3 [49] as “A UNI Service attribute in which more than 1331
one CE-VLAN ID can be associated with an EVC” and is described in Section 9.12/MEF 10.3 1332
[49]. All to one bundling defined as “A UNI attribute in which all CE-VLAN IDs are associated 1333
with a single EVC” is a special case of Bundling Enabled and is described in Section 9.13/MEF 1334
10.3 [49]. Table 12/MEF 10.3 [49] provides valid combinations for “Service multiplexing”, 1335
“Bundling” and “All to one bundling” attributes. 1336
1337
EPL must have the “Bundling” attribute as disabled and “All to one bundling” attribute as enabled. 1338
For EVPL the “Bundling” attribute can be enabled or disabled, but “All to one bundling” must be 1339
disabled. 1340
13.4.3 CE-VLAN ID preservation for EVC 1341
For EPL, CE-VLAN ID preservation must be enabled and CE-VLAN ID is preserved for EVC 1342
over the PSN. 1343
1344
For EVPL, if CE-VLAN ID preservation is enabled, CE-VLAN ID is preserved for EVC over the 1345
PSN. 1346
1347
Note: If CE-VLAN ID preservation is enabled, No VID translation is supported for the EVC. 1348
EVPL can support VID translation when using EVPN service type “VLAN-Based Service type”. 1349
13.4.4 CE-VLAN CoS preservation for EVC 1350
CE-VLAN CoS preservation service attribute defines whether the CE-VLAN CoS bits are 1351
preserved (unmodified) across the EVC. 1352
1353
For EPL, CE-VLAN CoS preservation may be enabled or disabled (see Table 12/MEF 6.2 [50]). 1354
For EVPL CE-VLAN CoS is supported as per sec 10.3/MEF 6.2 [50] and may be enabled or 1355
disabled. In an EVC with CE-VLAN CoS preservation is enabled, the EVPN preserves the CoS 1356
bits over PSN. 1357
13.4.5 EVC Maximum Service Frame Size 1358
The mapping from “EVC Maximum Service Frame Size” to “EVC MTU” is provided in Table 40 1359
Appendix B of MEF 6.2 [50]. 1360
1361
The EVC Maximum Service Frame Size is configurable with a default value of 1600 bytes. 1362
1363
When Ethernet frames are transported in MPLS networks, the MPLS packet includes the labels, 1364
and the EVC frame as payload. The path MTU is the largest packet size that can traverse this path 1365
without fragmentation. The ingress PE can use Path MTU Discovery to find the actual path MTU. 1366
1367
[R-82] PE SHOULD support configurable EVC Maximum Service Frame Size of at least 1368
1600 bytes (see Table 6/MEF 6.2 [50]). 1369
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 56 of 65
13.4.6 Frame delivery 1370
The frame delivery policy rules enable the service provider to specify how different frame types 1371
are handled by a PE. They enable setting specific rules for forwarding, discarding or conditionally 1372
forwarding specific frame types. The frame types used by the rules are: 1373
1374
• Unicast 1375
• Multicast 1376
• Broadcast 1377
1378
[R-83] PE MUST support setting policy function of frame delivery rules for forwarding, 1379
discarding or conditionally forwarding unicast, multicast and broadcast frames per EPL and 1380
EVPL services. 1381
13.4.7 Layer 2 control protocols 1382
The Layer 2 control protocol processing is independent of the EVC at the UNI. L2CP handling 1383
rules are set according to the definition of Section 8/MEF 6.1.1 [48] and differ per-service type. 1384
The PE policy function supports setting of rules for handling L2CP per service type. 1385
1386
EVC L2CP handling per service type can be set to: 1387
• Discard – Drop the frame. 1388
• Peer can be applicable: For example L2CP/LAMP, Link OAM, Port Authentication, and E-1389
LMI. 1390
• Tunnel – Pass to the egress UNI. 1391
1392
[R-84] PE MUST support policy function setting of rules for handing L2CP per service 1393
type as specified in Section 8/MEF 6.1.1 [48]. 1394
1395
Note: This specification only supports MEF 6.1.1 for L2CP processing equirements. 1396
Support of multiple-CEN L2CP MEF 45 [51] is outside the scope of this document. 1397
13.4.8 EVC performance 1398
The performance parameters indicate the quality of service for that service instance. They consist 1399
of the following: 1400
• Availability 1401
• Frame Delay 1402
• Frame delay variation 1403
• Frame loss ratio 1404
1405
The requirements for support of CoS and mapping are specified in QoS Section 9. 1406
1407
[R-85] The PE MUST support MEF SOAM performance monitoring as per MEF 35 [47]. 1408
1409
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 57 of 65
For transport of SOAM see Section 8.2. 1410
13.4.9 Multiple Class of Service 1411
MEF 6.1 [41] introduced services that were allowed multiple Classes of Service. The different 1412
ways that EVCs can be created to support such a service is defined in Section 11.4.9/MEF 6.1 [41]. 1413
1414
[R-86] The PE MUST support a single class of service for an EVC by using a single value 1415
for the TC bits in the MPLS label stack. 1416
1417
[R-87] The PE MUST support a multiple class of service for an EVC by using multiple 1418
values for the TC bits in the MPLS label stack. 1419
1420
1421
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 58 of 65
14 EVPN enabled Rooted-multipoint Ethernet VPN services (E-Tree) 1422
MEF Ethernet Tree (E-Tree) service type is based upon a Rooted-multipoint Ethernet Virtual 1423
connection. The EVPN technology enables the creation of E-Tree services over an MPLS network. 1424
EVPN can be used to create the EP-Tree and EVP-Tree services of the E-Tree service type defined 1425
by MEF 6.2 [50]. In an E-Tree service, endpoints are labeled as either Root or Leaf sites. Root 1426
sites can communicate with all other sites. Leaf sites can communicate with Root sites but not with 1427
other Leaf sites. 1428
1429
This service could be useful for internet access or video over IP application, such as 1430
multicast/broadcast packet video. 1431
1432
High-level services are described in Section 14.1and 14.2. In addition to the Carrier Ethernet 1433
defined service characteristics, EVPN significantly enhances important service characteristics such 1434
as reliability and scalability. The EVPN requirements for E-Tree Ethernet VPN services are listed 1435
in Section 14.3. 1436
Ethernet Private Tree (EP-Tree) 1437
The Ethernet Private Tree (EP-Tree) service uses a Rooted-multipoint EVC. In a multipoint EVC, 1438
two or more UNIs must be associated with one another. The EP-Tree service is defined to provide 1439
CE-VLAN tag preservation and tunneling of key Layer 2 control protocols. A key advantage of 1440
this service is that VLANs can be configured across the sites without any need to coordinate with 1441
the service provider. 1442
1443
Each interface is configured for “All to One Bundling”. EP-LAN supports CE-VLAN CoS 1444
preservation. Service multiplexing is disabled on the UNI. 1445
1446
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 59 of 65
1447 1448
Figure 8 Ethernet Private Tree (EP-Tree) Service 1449
Ethernet Virtual Private Tree (EVP-Tree) 1450
The Ethernet Virtual Private Tree (EVP-Tree) service allows service multiplexing at the UNI. It 1451
allows users of an E-Tree service to interconnect their UNIs and at the same time access other 1452
services (e.g. E-Line). Figure 9 shows an example of multiple services access from a single UNI. 1453
In this example, the user has an EVP-Tree service for Rooted-multipoint data connectivity and an 1454
EVPL service (P2P EVC) for accessing a value-add service from one of the UNIs. 1455
1456
Bundling can be used on the UNI in the EVP-Tree service and supports CE-VLAN tag 1457
preservation. All to One Bundling is disabled. 1458
1459
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 60 of 65
1460 1461
Figure 9 Ethernet Virtual Private Tree (EVP-Tree) Service 1462
1463
EVPN for establishing EP-Tree and EVP-Tree 1464
L2VPN service provides multipoint to multipoint connectivity for Ethernet across an IP or MPLS 1465
Network. A generic E-LAN/E-Tree service is always bidirectional in the sense that ingress frames 1466
can originate at any endpoint in the service. 1467
1468
The only difference between E-LAN and E-Tree is: 1469
• E-LAN has Root endpoints only, which implies there is no communication restriction 1470
between endpoints. 1471
• E-Tree has both Root and Leaf endpoints, which implies there is a need to enforce 1472
communication restriction between Leaf endpoints. 1473
1474
RFC 7387[38] proposes the solution framework for supporting E-Tree service in MPLS networks. 1475
The document identifies additional components to emulate E-Tree service using Ethernet LAN 1476
service over MPLS network. 1477
1478
EVPN provides support for the E-LAN service type in MPLS networks as described in Section 11. 1479
RFC 7432 [39] describes procedures for BGP MPLS-based Ethernet VPNS. EVPN requires 1480
extensions to existing IP/MPLS protocols. EVPN supports both provisioning and signaling for 1481
Ethernet VPNs and incorporates flexibility for service delivery over Layer 3 networks. 1482
1483
The PE supports BGP MPLS-based Ethernet VPN signaling and provisioning as specified in [R-1484
53] of Section 11.4. 1485
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 61 of 65
draft-ietf-bess-evpn-etree [52] describes how EVPN can be used to support E-Tree in IP/MPLS 1486
networks. The RFC discusses how the functional requirements for E-Tree service can be 1487
implemented efficiently with EVPN. 1488
14.3.1 E-Tree support in EVPN 1489
E-Tree Scenarios and EVPN Support 1490
The support for E-Tree services in EVPN is divided into different scenarios, depending on the site 1491
association (Root/Leaf) per PE or Access Ethernet Segment. They are described in section 2/ 1492
draft-ietf-bess-evpn-etree [52]. 1493
1494
[R-88] The PE routers MUST Support Leaf or Root site(s) per PE as defined in Section 1495
2.1/ draft-ietf-bess-evpn-etree [52]. 1496
1497
[R-89] The PE routers MUST Support Leaf or Root site(s) per Attachment Circuit (AC) as 1498
defined in Section 2.2/ draft-ietf-bess-evpn-etree [52]. 1499
1500
E-Tree Extended Community 1501
A new BGP extended community is used for leaf indication. They are described in section 5/ draft-1502
ietf-bess-evpn-etree [52]. 1503
1504
[R-90] The PE routers MUST Support BGP extended community encoding as defined in 1505
Section 5/ draft-ietf-bess-evpn-etree [52]. 1506
E-Tree Operation for EVPN 1507
RFC 7432 [39] has inherent capability to support E-Tree service. It adds a new 1508
BGP extended community for leaf indication. The E-Tree operations for EVPN are 1509
described in section 3/ draft-ietf-bess-evpn-etree [52]. 1510
[R-91] The PE routers MUST Support E-Tree operation for EVPN as defined in Section 3/ 1511
draft-ietf-bess-evpn-etree [52]. 1512
14.3.2 Service Interfaces 1513
EVPN supports several service connectivity options for delivering MEF services. They are 1514
provided in Section 11.2. MEF service requirements for EP-LAN and EVP-LAN are different. 1515
For EP-LAN, each interface is configured for “All to One Bundling”. For EVP-LAN “All to One 1516
Bundling” is disabled. 1517
Service Interfaces for EP-Tree 1518
[R-92] The PE routers MUST support Port-based Service Interface as defined in Section 1519
6.2.1/RFC 7432 [39]. 1520
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 62 of 65
1521
Service Interfaces for EVP-Tree 1522
[R-93] The PE routers MUST Support VLAN-based Service Interface as defined in Section 1523
6.1/RFC 7432 [39]. 1524
1525
[R-94] The PE routers MUST support VLAN Bundle Service Interface as defined in 1526
Section 6.2/RFC 7432 [39] 1527
14.3.3 Data plane 1528
The requirements for data plane per Section 12.3.2 are applicable. 1529
14.3.4 Tunnel signaling 1530
The requirements for tunnel signaling per Section 12.3.3 are applicable. 1531
14.3.5 Routing 1532
The requirements for routing per Section 7.2 are applicable. 1533
14.3.6 Multi-Homing and Load balancing 1534
The requirements for multi-homing and load balancing per Section 12.3.5 are applicable. 1535
14.3.7 OAM 1536
The requirements for OAM per Section 12.3.6 are applicable. 1537
14.3.8 Convergence 1538
Failure recovery from different types of network failure is supported as per Section 8.3.2. 1539
14.3.9 PSN Resiliency 1540
PSN resiliency is supported as per Section 10. 1541
1542
Fast Convergence 1543
Fast convergence is supported as per Section 11.6. 1544
14.3.10 Multicast and Broadcast 1545
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 63 of 65
[R-95] PE routers SHOULD support multicast and broadcast traffic as per Section 16/RFC 1546
7432 [39]. 1547
14.3.11 QoS 1548
In general, an E-Tree service type can provide a best effort service with no performance assurance. 1549
In certain cases, an E-Tree service type can be defined with performance objectives (see Section 1550
9.2/MEF 6.2 [50]. 1551
1552
[R-96] PE routers SHOULD support the QoS mapping as per Section 9. 1553
14.3.12 Security 1554
[R-97] PE routers MUST support security as per Section 19/RFC 7432 [39]. 1555
1556
Support of service attributes for EP-Tree and EVP-Tree 1557
Section 9.3/MEF 6.2 [50] specifies the E-Tree service type. Section 10.5 and 10.6/MEF 6.2 [50] 1558
provides service attributes and parameters for EP-Tree and EVP-Tree services respectively. TR-1559
224 refers to MEF 6.1 [41] and MEF 10.2. TR-350 uses the backward compatible subset of the 1560
revised specifications MEF 6.2 [50] (see Appendix B) and MEF 10.3 [49] to achieve the 1561
equivalent function. 1562
1563
Some of the service attributes and parameters are provided by Ethernet physical interface and 1564
service provisioning (e.g., Physical medium, Speed, Mode, MAC layer, EVC type, maximum 1565
number of EVCs, etc.). This section only describes those service attributes and parameters that are 1566
relevant to transporting the EVPN traffic over PSN. 1567
14.4.1 Bandwidth Profile 1568
A bandwidth profile defines how rate enforcement of Ethernet frames is applied at an UNI. 1569
Bandwidth profiles enable offering service bandwidth below the UNI access speed (aka Speed) 1570
and limit the amount of traffic entering the network per the terms of the SLA. 1571
1572
For Tree services, bandwidth profiles can be optionally specified per-UNI (ingress and egress), 1573
per-EVC (ingress and egress), and/or per CoS (ingress and egress). An E-Tree service can be 1574
provided as a best effort service without any bandwidth guarantee. 1575
1576
Note: Bandwidth profiles can be optionally provisioned. They are not carried by EVPN protocols. 1577
1578
1579
[R-98] A PE SHOULD support the bandwidth profile algorithm as per the portion of 1580
Section 12/MEF 10.3 [49] that is backward compatible with MEF 10.2 [42]. 1581
1582
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 64 of 65
In order to support bandwidth profile, technique such as admission control and Diffserv-TE as 1583
specified in Section 9 are used. 1584
14.4.2 Bundling 1585
Section 9.12/MEF 10.3 [49] specifies the bundling service attribute. Bundling implies “A UNI 1586
attribute in which more than one CE-VLAN ID can be associated with an EVC”. All to one 1587
bundling enabled is a special case of bundling. It implies “A UNI attribute in which all CE-VLAN 1588
IDs are associated with a single EVC”. Table 12/MEF 10.3 [49] provides valid combinations for 1589
“All to one bundling” and “Service multiplexing” attributes. 1590
1591
EP-Tree must have “All to one bundling” attribute enabled. For EVP-Tree the bundling attribute 1592
can be enabled or disabled. However, for EVP-Tree “All to one bundling” must be disabled. 1593
14.4.3 CE-VLAN ID preservation for EVC 1594
CE-VLAN ID preservation service attribute defines whether the CE-VLAN ID is preserved 1595
(unmodified) across the EVC. 1596
1597
For EP-Tree, CE-VLAN ID preservation must be enabled and CE-VLAN ID is preserved for EVC 1598
over the PSN. 1599
1600
For EVP-Tree, if CE-VLAN ID preservation is enabled, CE-VLAN ID is preserved for EVC over 1601
the PSN. 1602
1603
Note: If CE-VLAN ID preservation is enabled, No VID translation is supported for the EVC. 1604
EVP-Tree can support VID translation when using EVPN service type “VLAN-Based Service 1605
type”. 1606
14.4.4 CE-VLAN CoS preservation for EVC 1607
CE-VLAN CoS preservation service attribute defines whether the CE-VLAN CoS bits are 1608
preserved (unmodified) across the EVC. 1609
1610
For EP-Tree, CE-VLAN CoS preservation must be enabled (see Table 21/MEF 6.2 [50]) and CE-1611
VLAN CoS is preserved for EVC over PSN. 1612
1613
For EVP-Tree, CE-VLAN CoS preservation can be either enabled or disabled (see Table 24/MEF 1614
6.2). In an EVC with CE-VLAN CoS preservation is enabled, the EVPN preserves the CoS bits 1615
over PSN. 1616
14.4.5 EVC Maximum Service Frame Size 1617
The requirements for EVC maximum service frame size per Section 12.4.5 are applicable. 1618
Ethernet Services Using BGP MPLS based EVPNs TR-350 Issue 2
TBD 2017 © The Broadband Forum. All rights reserved 65 of 65
14.4.6 Frame delivery 1619
The frame delivery policy rules enable the service provider to specify how different frame types 1620
are handled by a PE. They enable setting specific rules for forwarding, discarding or conditionally 1621
forwarding specific frame types. The frame types used by the rules are: 1622
1623
• Unicast 1624
• Multicast 1625
• Broadcast 1626
1627
[R-99] PE MUST support setting policy function of frame delivery rules for 1628
forwarding, discarding or conditionally forwarding unicast, multicast and 1629
broadcast frames per EP-Tree and EVP-Tree services. 1630
14.4.7 Layer 2 control protocols 1631
The requirements for Layer 2 control protocols per Section 12.4.7 are applicable. 1632
1633
14.4.8 EVC performance 1634
The requirements for EVC performance per Section 12.4.8 are applicable. 1635
1636
1637
1638
1639
1640
End of Broadband Forum Technical Report TR-350 1641
1642
1643
1644
1645